FR3097149A1 - Electronic control device optimized for specific mobile robot - Google Patents

Abstract

Electronic control device optimized for a specific mobile robot The device according to the invention uses at least one robotic arm (1), carried by a mobile base (2) (remote controlled, autonomous, wire-guided, etc.). The device according to the invention also implements control software (300), algorithms (320) operating in parallel, and a specific configuration library (301). The device according to the invention is particularly intended for the production process of physical goods, and for the transport of goods (including palletizing). The device is also suitable for mobility on unstructured terrain. Figure to be published with the abstract: Fig. 1

Description

Electronic control device optimized for specific mobile robot

The present invention relates to a device for grasping, carrying, transporting and then depositing more or less heavy loads, with possibly high precision. On the other hand, this device can also exert a force on the load carried, and/or on an element of the environment. The device according to the invention implements at least one arm carried by a mobile platform (for example an autonomous, remote-controlled, or guided robot, etc.) whose action is located, for example:

within a production process by participating in the flow of components, their development or their assembly,

upstream or downstream of a production process, by bringing together or dissociating components, which are identical or different from each other, to prepare for transport.

The device according to the invention is particularly suitable for handling objects, with a mass of 2 to 500 kg, of unconstrained size, possibly with great precision, whether in a usual environment (mechanical production, handling of packages, preparation control in logistics) or in a severe and/or unstructured environment such as: agriculture, underground, woods and undergrowth, military, indoor or outdoor urban, and post-earthquake urban..., for example: refrigeration temperature (food industry), high temperature (metallurgical process), hazardous (construction, chemicals, oil, military, civil security), controlled environment with low dust level (electronic production), controlled environment in microbiology with a high level of asepsis (health, vaccine); for activities in the usual domain (distribution, automobile) or sensitive (handling of banknotes, valuables, medicines, weapons, etc.).

The device according to the invention is also indicated for the traditional use cases of mobile robots described in the following chapter: “Context”.

Context

First cases of traditional use: When the handling of a heavy load (for example boxes of goods, mechanical parts, tools, cylinders) is of unpredictable, irregular, infrequent and/or complex frequency, then the handling is traditionally carried out by a human operator. This manual handling presents ergonomic constraints for the workstation (storage at head height, ball table, etc.) and posture constraints for the operators (see hardship). It is also regulated with a maximum permissible mass of around 10kg (variable limit depending on the sector of activity, and cumulative daily, weekly, etc.). Finally, manual handling involves a long-term health risk for operators, namely musculoskeletal disorders (MSDs). A first level of automation is the use of tools, such as hoists and winches, which are restrictive for the safety of operators (because the heavy parts are moving at the end of a chain, therefore with imprecise guidance) and cumbersome in terms of installation at height and on the ground, and these tools often do not completely eliminate the ergonomics and MSD risk for operators.

A particular case of interest for the invention is the logistics upstream of a production process, a retail food outlet, a pharmacy, or any establishment having the need to constantly present a wide range of products. different with a small amount of each. This upstream order preparation activity (distribution, or breakdown) is traditionally carried out manually, which limits its use to products with higher added value (such as drugs). Alternatively, the preparation of orders can be carried out or facilitated, by automated storage towers, carousels or continuous flow sorting processes (for example: sorting of mail by La Poste), but the significant investment is only justified in the case of intensive use, and only if all the preparation combinations are known and validated in advance. So today, order preparation relies on human manipulation.

Second traditional use case: In stable and large series production processes, heavy loads are usually handled by a massive robotic arm fixed to the ground: it is usual to have a ratio of 10 to 30% between the mass of the arm and that of the object handled (e.g.: the 960kg EverRobot MPL80 arm handles an 80kg load at 2.2m from its axis).

In addition, a zone reserved for the movement of the robotic arm with its load, protected by a barrier, must be installed. And the automation by robotic arm of the manipulation of a part (or other) between two machines, imposes to arrange the machine from which the part comes out, and the one where it goes, in a way adapted to the robotic arm, in particular its elongation , its dexterity (number of degrees of freedom defining its ability to move) and to take into account the accessibility zones of the machines. Finally, the machines and the arm must be managed in a coordinated way, which requires that the arm and the machines be equipped with sensors allowing at least collisions between them to be avoided. This secure, controlled and specialized area is a robotized production cell.

As it is necessary to leave passage spaces between the machines and the robotic arm for the operators, in particular for maintenance, the size of a robotic cell is generally several meters on each side. The machines in a cell are therefore usually 3 m apart, which requires the arm to have a range of 2 m on either side (0.5 to enter the machine to pick up the part). The design of the typical cell therefore results in a need for an arm capable of handling a heavy load at a distance of 2m, hence the use of massive robots, equipped with very large motors, which consume and heat a lot, d a large load-bearing structure, which is expensive to purchase, complex to install and maintain, and which generate significant inertia, which is dangerous for operators, hence the protected area. Finally, the dangerousness of the parts carried by the robotic arms with a fixed base is all the more important as the users maximize productivity, by increasing the speed of the movements, hence a high kinetic energy.

As illustrated above, the implementation of robotic cells imposes strong constraints on the flow of parts, the layout of the machines, the high footprint, as well as a great technicality in the installation, the adjustment and the maintenance, a significant investment in the purchase and use of these massive robotic arms, and significant danger.

Third traditional use case: for a multitude of reasons, it happens that the seizure of a part or its manipulation in space requires great precision. These reasons can be that the part is partially soft, temporarily weakened by the production process, or that the part is piled up in bulk with several others in the same packaging, or the part can be of a complex shape which imposes to grasp it very precisely. The precision of a robotic arm requires a certain type of sensor and effector, but above all, this requires that the mechanical structure be very rigid, which traditionally results in making the robotic arm heavier, again inducing the aforementioned constraints.

Fourth traditional use case: each industrial operation is justified by a demonstrable economic return. However, to make the significant investment profitable, the robots must show high productivity, hence a fast working speed. Responding to this justified economic requirement further increases the stress on robotic arms, because not only is it necessary to handle remote loads, possibly heavy, possibly with precision, but it must be done quickly. Traditionally, the dynamic capacities of the motors are then greatly increased, resulting in an additional weighting of the structure, with possibly a lowering of the capacity to handle heavy loads from a distance.

Technical background and pre-existing patents

There is a very rich bibliography, and technical achievements in the field of process automation, robotics and mobile robotics, for industrial uses, mainly for logistics or handling, as well as in unstructured environments.

The numerous patents of the ABB and Kuka companies are not cited here, because they focus on distant themes of the invention (e.g.: problem of thermal heating for heavy robot arms, data fusion, or on control loop drift correction aspects for lighter and more precise arms).

The pre-existing patents in relation to the present invention are:

FR1901856, from HMS2030 (filed 02/22/2019): which notably claims joint locking devices. These devices are named in the present invention: "dedicated mechanism (190)", and are part of the hardware elements that can increase the performance of the robot by integrating into the device described in the present invention.

US9499219, from Google (2016) claims a method for detecting contact between a robotic limb and the ground, based on a mechanical-pneumatic system, and presents associated algorithms.

US9393693, from Google (2016) claims a robotic manipulator system including a gripper, and a method for generating a virtual space of constraints relating to a set of force vectors, limiting the acceptable trajectories in order to avoid the fall of the carried part . The described method seems relevant only for the elimination of inadequate trajectories of the carried object. But it remains passive, namely it does not generate any solution, and on the other hand the claims suggest significant shortcomings for the control of a mobile robot, for example its balance, the taking into account of its motor capacities and its limitations. angular joints... Furthermore, the proposed use of artificial intelligence (particle filter) is a compromise between optimized execution and minimization of computing power. Nevertheless, the computing power required to run artificial intelligence software remains high, resulting in a reaction time that may be too long in the event of a rapid change in the robot's environment (risk of falling, etc.).

WO2009/143377 by Georgia Tech Research Corp: claims a robot with one or two wheels, in particular a humanoid, whose static and quasi-static stability is ensured by the movement of a pendulum system, namely typically an arm of the robot.

US2016/0229057 by Harris Corp: claims a method for avoiding the overturning of a robotic vehicle (on sloping ground) by using a mobile component carried by the robot, the position of which can be subsequently optimized. This patent deals with the control of the balance of a robot arm on a mobile base on an inclined terrain, and this balance is achieved only by moving a mobile object carried by the robot; that is to say that this patent does not consider the possibilities of adaptation to the terrain of the robot itself.

US2001/0045807 inventor Mc Connell: claims a method for controlling the movement of an object (carried by a robotic arm) based on the measurement of the torque and the position, allowing in particular to eliminate the oscillations of a robot arm.

It appears that the state of the art only addresses the control of mobile robots and manipulators in a piecemeal way. Even the implementation of all the sensor devices, and control methods, described in the state of the art, would not provide a coherent software and hardware set, the operation of which would make it possible to meet the constraints of a actual use.

Moreover, since mobile robots offer thousands of possibilities and redundant geometries to pick up the same part, or to move from point A to point B. The computing power required to evaluate these options is therefore very high. Very few patents address this point, and none with comparable effectiveness to the present invention.

Advantages of the invention

The invention describes specific equipment, and a dedicated control method (and vice versa). The system thus made up of equipment & control method, can perform various actions in realistic environments, in an optimal way, by implementing a minimum of resources.

The method is based on an initial library containing a limited number of possible and relevant configurations, for example between 10 and 80, and typically 50. New configurations are developed, at each iteration of the algorithms (320) by these same algorithms, by taking take into account the actual execution conditions of the assigned action, between 5 and 50, and typically 20 variants per initial configuration (beyond the first iteration, the configurations with the best scores during the previous iterations, can be preferred for the generation of variants). The computing power is therefore concentrated on a small amount of information, namely typically 1000 configurations per calculation iteration. And it is between 50 and 800, typically 200 iterations that are made, to avoid sub-optimal solutions. The computing power required according to the invention is 10,000 times lower than existing approaches (vs. AI-particle filter, see Google patent). This software efficiency offers the possibility of adapting the configuration of the robot several times per second during the execution of the action! This responsiveness is synonymous with reliability. In addition, by relaxing the constraint of computing power, it becomes possible to use proven, compact and less expensive electronic cards.

The system according to the invention evaluates the feasibility of an action, and can increase the capabilities of the robot, for example by generating sub-objectives that are less restrictive than the action as assigned, or by limiting the opening angle of a joint in order to limit the lever arm exerted on its motor, or by blocking a joint thus making it possible to exceed the maximum torque of the motor concerned, or by leaning on an element of the environment to stabilize itself. The system according to the invention also makes it possible to optimize the execution of the assigned action, according to selected criteria, such as current consumption, execution speed, etc.

Glossary

“Target” is a sub-objective of the assigned action, an assigned action being at least made up of a target, and typically made up of a starting target and a finishing target; the achievement (or execution, or achievement) of all targets, and transitions between targets, is the execution of the complete assigned action. Each target is provided or described in a complementary way between the human, the supervision software, and / or the robot sensors (3), so that a target is always described exhaustively. The characteristic data of a target are typically:

its logical order in the execution of the assigned action, called sequencing

the spatial coordinates tolerated for each point, or each geometric element, of the environment (6), targeted, and where the robot (3) must position at least one of its components, typically the gripper

the spatial coordinates tolerated for each point, or geometric element, of the environment (6) constituting a constraint or an obstacle, and where the robot (3) must not position any of its components,

"Configuration" defines: the position in space of the base (2), the geometry of the arm (1) relative to the base (2), the static and dynamic characteristics, namely speed and acceleration, of the components of the robot ( 3) and the object (4) possibly worn, and/or status characteristics such as motor consumption, their temperature, etc.; and finally each configuration can be added one or more fields containing, for example, an evaluation score for each target considered; this note characterizes the adequacy of said configuration to said target, this field can also contain logical information, in the form of text or other (e.g.: boolean), in order to allow the algorithms (320) to operate in an optimal manner .

"Rest configuration" corresponds to a configuration where the level of effort undergone by the robot (3) is either negligible, or does not induce a significant limitation of its capacities either of mobility or of manipulation.

1. l’angle (alpha_j_i) est l’angle formé, dans la configuration (j), entre le segment (110_i) et la gravité ;
2. (sin-1) est la fonction inverse du sinus,
3. (Cmax_i) est le couple de sortie du moteur (M_i), considéré comme « maximal » au vu de l’application
4. (Cres90_j_i) est le couple résultant maximal exercé sur l’axe du moteur (M_i), à savoir la somme des moments exercés sur le segment (110_i) ajouté à la somme des moments maximum des forces appliquées sur ce segment, à savoir quand ce segment forme un angle de 90° avec la gravité,

Nota Bene : si (Cmax_i/ Cres90_j_i) est supérieur à « 1 », alors le moteur (M_i) peut exécuter l’action sans contrainte angulaire ;
Application numérique: Nous avons 2 forces extérieures appliquées sur un segment du bras, son propre poids (0.25,1.8) et le poids de l’objet porté (0.6,50), où (x,y) correspond à : « x » étant la distance en (m) entre le point d’application de la force et l’axe du moteur, et « y » étant la masse en (kg). Il en résulte un couple maximal généré sur l’axe du moteur (Cres90_j_i) égal à 304.5 Nm. Considérons un couple moteur max de 160 Nm (Cmax_i), l’angle maximal de motricité (alpha_j_i) est de : 31.7°.In the motricity threshold equation, which defines the angle beyond which the motor is no longer capable of generating a movement of the segment it controls, and in the very frequent particular case, where the forces of gravity are preponderant with respect to other external forces, we have:
(alpha_j_i) <= (sin-1)(Cmax_i/Cres90_j_i)

1. the angle (alpha_j_i) is the angle formed, in the configuration (j), between the segment (110_i) and gravity;
2. (sin-1) is the inverse function of sine,
3. (Cmax_i) is the motor output torque (M_i), considered as "maximum" in view of the application
4. (Cres90_j_i) is the maximum resulting torque exerted on the axis of the motor (M_i), namely the sum of the moments exerted on the segment (110_i) added to the sum of the maximum moments of the forces applied on this segment, namely when this segment forms a 90° angle with gravity,

Please note: if (Cmax_i/ Cres90_j_i) is greater than "1", then the motor (M_i) can perform the action without angular constraint;
Numerical application: We have 2 external forces applied to a segment of the arm, its own weight (0.25,1.8) and the weight of the object carried (0.6,50), where (x,y) corresponds to: "x" being the distance in (m) between the point of application of the force and the axis of the motor, and "y" being the mass in (kg). This results in a maximum torque generated on the axis of the motor (Cres90_j_i) equal to 304.5 Nm. Let us consider a maximum motor torque of 160 Nm (Cmax_i), the maximum traction angle (alpha_j_i) is: 31.7°.

Description of the invention

The device is a robot (3), which comprises according to a first characteristic at least one robotic arm (1) carried by a mobile base (2), which allows the robot to move usually by carrying a part between a starting point, and an end point.

According to a second characteristic, the robotic arm is equipped with proprioceptive sensors (140), making it possible to measure its own position in space, as well as the torque and/or the forces which are exerted on its structure and its joints, and therefore its engines.

According to a third feature, the robotic arm and/or its mobile base is equipped with exteroceptive sensors (140) (240), making it possible to measure, for example: the dimensional, position, mass, possibility of gripping characteristics of the part to be handle, and the environment of the robot, in particular the ground and the presence of obstacles, authorizing its use without the need for a protected area and/or without a priori or certain knowledge of its environment (6).

According to a fourth feature, one or more electronic device (310), whether onboard or not, manages the motors (150) (250) of the robotic arm (1) and of its mobile base (2) in a coordinated manner. This electronic device (310) implements algorithms (320), and makes it possible to interpret the action assigned in a logical and mathematical form, and also to generate a set of sub-objectives, called "targets", necessary and sufficient to the execution of the assigned action, and/or to complete with additional data, coming from sensor or calculation, the partial description of these targets possibly provided by the human and/or the supervision system.

According to a fifth characteristic, the arm (1) being made up of at least one movable segment (110), each segment (110) being moved by at least one motor (150); the segments (110) being interconnected by a hinge (180); the arm (1) being provided with at least one sensor (140); the arm (1) being fixed by one of its ends to the base (2) by a hinge (280); the arm (1) being able to integrate at the other end at least one gripper (120) being able to be equipped with at least one motor (150);

According to a sixth characteristic, the base (2) being equipped with at least one sensor (240) and at least one motor (250); the base (2) being mobile and provided with at least one contact member (220) with the ground, of the type: wheel, caterpillar, articulated leg, crutch or a combination of these elements

According to a seventh characteristic, the electronic device(s) (310) controlling, directly or indirectly, all of the motors (150) and (250) of the robot (3) in a coordinated manner; the electronic device(s) (310) being equipped with at least one memory (311) for storing information, and/or at least one means of communication (312) with one or more local computer networks , and/or at least one man-machine interface (313);

According to an eighth characteristic, the memory or memories (311) storing a set of data (333) resulting from the measurements carried out by the sensors (140) and (240), communicated via the communication means (312) or via the man-machine interface (313), or calculated by the control software(s) (300);

According to a ninth characteristic, the data (333) typically consisting of: the logical description of the action to be performed, the characteristics of the possible object (4), of the possible external element (5), and of the environment (6), in particular the ground, and the obstacles, and also the mechanical and functional characteristics of the robot (3), and finally one, or more, possible safety factor making it possible to overcome the uncertainties of measurement or calculation, of the data (333).

1. uniquement de configurations dites possibles, à savoir qui respectent les limitations mécaniques et motrices, en statique et dynamique, intrinsèques du robot (3),
2. initiallement d’une collection quasi-exhaustive des configurations dites pertinentes, à savoir qui utilisent les capacités motrices et manipulatoires du robot (3) à leur maximum,
3. de configurations possibles issues des configurations précitées, élaborées par le logiciel de commande (300) en vue de l’exécution de l’action assignée.

According to a tenth characteristic, the control software(s) (300) implementing a single configuration library (301), consisting of the control software(s) (300) implementing a single configuration library (301 ), consisting of:

1. only so-called possible configurations, i.e. which respect the mechanical and motor limitations, in static and dynamic, intrinsic to the robot (3),
2. initially of a quasi-exhaustive collection of so-called relevant configurations, namely which use the motor and manipulative capacities of the robot (3) to their maximum,
3. of possible configurations resulting from the aforementioned configurations, elaborated by the control software (300) with a view to the execution of the assigned action.

These configurations being provided during the manufacture, installation and/or operation of the robot (3), and/or generated by the control software (300),

According to an eleventh characteristic Device according to at least one of the preceding claims, characterized in that the control software (300) includes and implements algorithms (320), which operate in parallel, and which use iteratively as incoming data, between other results generated by the algorithms (320) during the present or previous iterations; the algorithms (320) provide the control software (300) with a set of information necessary and sufficient for the optimal execution of the assigned action; the aforementioned set of information is obtained by performing all or part of the following calculations, for each of the targets of the assigned action, the calculations from (i) to (iv) are executed in parallel, then (v) and finally ( vi) are executed sequentially or in parallel:

synthesize a substantially comprehensive set of constraints relating to the execution of each target

develop new possible configurations from the previously evaluated configurations, and adapt them to all the execution constraints mentioned in (i),

possibly generate one or more sub-objectives to be integrated into the assigned action, in order to be able to execute it, or in order to optimize its execution,

possibly identify in the environment (6) the element or elements on which the robot (3) can exert an effort,

evaluate the capacity of the robot (3) in each configuration considered, taken from the library (301), to satisfy all the constraints mentioned in (i), and calculate a score representative of this level of satisfaction,

compare and classify the configurations satisfying all the execution constraints mentioned in (i), for each target of the assigned action, to define, if it exists, the sequence of optimal configurations allowing to execute the complete assigned action , then validate this sequence.

According to a twelfth characteristic, among the algorithms (320) there is an algorithm (322) which, for each target individually, evaluates then selects the configurations of the library (301), which satisfy the conditions below (an algorithm variant (322) being that where one, or more, of these conditions is considered to be validated a priori, thus dispensing with checking it); the conditions are as follows:

if the robot (3) is able to reach the target respecting all the constraints of the data (333) then the configuration is said to be plausible,

if the sum of the forces external to the robot (3), namely the forces and the moments in the 3 dimensions, in static or quasi-static, resulting from the execution of the assigned action, is almost zero, or alternatively if the resulting acceleration remains under the control of the robot (3), i.e. can be canceled without falling or overturning, then the configuration is said to be balanced,

if the robot (3) is able to generate efforts greater than the efforts necessary to perform the assigned action, then the configuration is said to be efficient,

if the robot (3) is able to generate forces greater than the forces required for the two transitions below, namely firstly from a rest configuration, or from the configuration of the previous target (n-1) if it exists, to the configuration of the target (n) currently considered, and secondly from the configuration of the target (n) to the rest configuration, or to that of the next target (n+1) if it exists; then the configuration is said to be compatible,

All of these 4 conditions define the “synthetic constraints” for the target considered. The synthetic constraints are stored in the data (333). Configurations fulfilling all the conditions are said to be satisfactory.

According to a thirteenth feature, the electronic device(s) has the ability to compare the different satisfactory configurations, by assigning them a score. This score is established according to a calculation method provided by the human and/or the supervision system, or provided by the electronic device according to a predefined meta-rule, making it possible to generate the most relevant comparative function according to the data available to the device. electronic. The calculation method responds to an optimization problem based on a certain number of criteria, and a mathematical reprocessing of the data resulting from the simulation making it possible to classify the configurations according to their level of performance, and therefore to designate the most suitable for each target. individually.

According to a fourteenth characteristic, the electronic device(s) stores in memory all the possible configurations, as well as all the data necessary for the operation of the robot and the execution of the assigned action.

According to a fifteenth characteristic, among the algorithms (320) there is an algorithm (324) generating for each target, from configurations (j) not necessarily satisfying the synthetic constraints, of the library (301) potentially improved variants (j '); these configurations (j') are stored in the library (301); the algorithm(s) (324) using at least one of the following five configuration improvement methods:

bringing the base (2) closer to the target, while remaining beyond the minimum possible, defined by “D_environ_min”, which is the minimum possible distance authorized by the environment (6) between the joint (280) and the target; this displacement makes it possible to relax the equilibrium conditions, and to reduce the moments exerted on the components of the robot (3);

block one or more joints by using the brake integrated into the motor (150) (250), and/or with the dedicated mechanism (190), so as to be able to exert on the joint(s) concerned, the necessary force during the execution of the action, this force possibly being greater than that which the motor would provide if it were in operation

using at least one arm (1) to bring it into contact with an element of the environment (6) and to exert a force on said element; typically support on the ground improves stability, or traction with the arm (1) on a fixed element of the environment (6) contributes to bring the base (2) closer to this fixed element, or finally an obstacle can be pushed aside by the arm (1) or slide along it so that it does not interfere with the movement of the robot,

maintain, for each motor, the torque required to perform the rated action lower than the maximum admissible torque; this method is applicable for any articulation (180) (280) of index (i) having a degree of freedom in rotation, driven by a motor (150) (250) called (M_i); this method can transform an initially unsatisfactory configuration (j) into an efficient configuration (j')
in the very frequent special case, where the forces of gravity are preponderant compared to the other external forces, this method consists in limiting the angle "alpha_j_i" below the motor's angular traction threshold (M_i), according to the equation:
(alpha_j_i) <= (sin-1)(Cmax_i/Cres90_j_i)
this method can be implemented typically by successively adapting the angles (alpha_j_i) of each segment (110_i) starting with that linked to the joint (280);
Please note: if (Cmax_i/ Cres90_j_i) is greater than "1", then the motor (M_i) can perform the action without angular constraint;

smooth or equalize certain parameters of certain components of the robot (3) by adapting the angles (alpha_j_i), these parameters typically being the consumption or the torques of the motors (M_i), or the mechanical stresses undergone by the joints (180) (280) ; when these parameters are criteria to be optimized, then this method makes it possible to optimize the configuration (j)

According to a sixteenth characteristic, among the algorithms (320) there is an algorithm (325) making it possible to add at least one new target (n'), called intermediate, after the execution of the target (n) and before the target ( n+1); this addition takes place in two cases:

when the algorithm (322) has identified for each target, at least one configuration satisfying all the conditions except "compatibility", then the addition of a dedicated intermediate target (n') allows the transition of the incompatible configuration of the target (n) towards the configuration necessary for the next target (n+1); this intermediate target (n') typically being the placing of the object (4) on an element of the environment (6), for example the ground, a table, or a suitable location for the robot (3), typically its platform, to then transport it, and/or re-enter the object (4) according to a configuration compatible with that of the target (n+1)

when among the algorithms (320) there is an algorithm (326), indicating that the transition from the target (n) to the next one (n+1) induces a significant modification of the synthetic constraints compared to those of each of these successive targets , then the addition of a target (n') allows an adaptation of the configuration to these different constraints during the transition; it may in particular be a relaxation of stresses allowing the arm (1) to deposit the object (4), or a hazard such as the presence of a fixed or mobile obstacle, or even a hardening equilibrium conditions such as sloping or slippery ground.

According to a seventeenth feature, among the algorithms (320) is an algorithm (323) which firstly notes individually, for each target, all the configurations studied by the algorithm(s) (322), and on the other hand assigns an overall mark for each possible succession of satisfactory configurations making it possible to carry out the complete action; the notes are stored in memory (311). Ratings are based on one or more criteria to be optimized, weighted and/or statistically restated; this, or these, criterion, and statistical weighting or reprocessing rule being supplied to the robot (3) and/or generated by the control software (300).
Then the algorithm (323) compares the scores on the present cycle (or iteration) of calculation of the algorithms (320) with those obtained previously and stored in memory (311). For each target, the configuration with the best score (of all the configurations of the present cycle, and of all the cycles) is the so-called optimal configuration for said target; and for the complete action the succession having the best score (of all the successions of the present cycle, and of all the cycles) is the so-called optimal succession.
And finally before launching the optimal succession allowing the execution of the action, the algorithm (323) will have to confirm the validity of said optimal succession by using one or more validation criteria, for example: a minimum threshold of iterations of the algorithms (320), for example 200 cycles, and/or a threshold value above or below which the overall score, or alternatively the scores for each configuration of the succession, must be found to be validated...

According to an eighteenth characteristic, the configuration library (301) contains in particular a quasi-exhaustive collection of so-called relevant configurations, namely which use the motor and manipulative capacities of the robot (3) to their maximum.

According to a nineteenth characteristic, the control software(s) (300) implements the set of algorithms (320) once when assigning the action, in order to determine an embodiment according to the data (333) known at the start, and also regularly and closely over time, throughout the action, in order to enable its actual execution and to optimize its progress.

According to a twentieth feature, where the algorithms (320), part of the control software(s) (300), the configuration library (301) and part of the data (333) are remote, namely on a fixed PC or equivalent, typically a high-powered server, and which communicates via the local computer network with the robot (3), making it possible to reduce the electronic device (310) on board the robot (3) to what is strictly necessary, and therefore reducing the cost of the robot (3), while maintaining high computing power making it possible to maintain or even increase the responsiveness of the robot (3).

According to a twenty-first characteristic: among the algorithms (320) there is an algorithm (321) verifying that there exists for each target at least one satisfactory configuration, thus allowing by at least a succession of satisfactory configurations to carry out the complete assigned action; and as long as there does not exist for each target at least one satisfactory configuration, then the algorithms (320) reiterate their calculation loop (or cycle or iteration) from a completed configuration library (301), and d a possibly larger target set.

According to a twenty-second characteristic: among the algorithms (320) there is an algorithm (327), which studies the configuration of the robot (3) being calculated for each target, and is activated in the event that at least one arm (1) is available to exert a force on an element of the environment (6); then said algorithm (327) identifies in the environment (6) the element(s) on which the arm (1) can exert an effort, in order to increase the performance and/or the stability of the robot (3). For each new element available in the environment (6) at least one new configuration will be generated by the algorithm (324), with a view to an evaluation of the new configuration(s) by the algorithm (323).

According to a twenty-third characteristic: among the algorithms (320) is the algorithm (324) which preferentially generates, for each target individually, at least one configuration (j'), from one, or more, configuration (j) from the library (301) selected according to one or more criteria relating to the score assigned by the algorithm (323) during a previous iteration of the algorithms (320), the criterion relating to the score possibly typically be: the configuration (j) having obtained the best score, or the configurations (j) having obtained a score above a certain threshold, or else the "X" best configurations (j) (for example: X=5). ..

According to a preferred embodiment, the mobile base has a substantially rectangular shape, is equipped with 3 or 4 wheels, and where one, or two, arms are fixed relatively close to one side of the rectangle, and where:

• est sensiblement horizontale
• permet de porter un contenant de transport (710), le bras (1) pouvant venir déposer au moins un objet (4) sur ladite plateforme, ou sur ledit contenant de transport (710), ce qui permet de déplacer un, ou plusieurs, objet de manière stable et sans solliciter mécaniquement le bras, minimisant donc son usure et sa consommation d’énergie (cf configuration de repos)
• a une forme adaptée au contenant de transport (710), typiquement une palette de transport logistique de format standard européen ou US, la forme de la plateforme (7) étant alors typiquement une fourche de dimensions identique à celles d’un chariot transpalettes usuels
• est éventuellement mobile selon un axe vertical, permettant ainsi notamment de soulever du sol, respectivement déposer, un contenant de transport (710),

a platform (7) is arranged on the base (2), characterized in that it:

• is substantially horizontal
• makes it possible to carry a transport container (710), the arm (1) being able to deposit at least one object (4) on said platform, or on said transport container (710), which makes it possible to move one, or more, object in a stable way and without mechanically stressing the arm, thus minimizing its wear and its energy consumption (cf rest configuration)
• has a shape adapted to the transport container (710), typically a logistics transport pallet of European or US standard format, the shape of the platform (7) then being typically a fork of dimensions identical to those of a usual pallet truck
• is optionally movable along a vertical axis, thus making it possible in particular to lift from the ground, respectively deposit, a transport container (710),

the arm(s) consisting of 2 to 5 segments (110) and at least one gripper (120)

at least one of the robot's motors (3) is equipped with an internal brake, integrated into the motor or arranged a posteriori, and/or at least one dedicated mechanism (190) for blocking the degree of freedom authorized by said motor

a counterweight (290), removable or fixed, is possibly present, and is typically fixed on the side of the base (2) opposite the arm (1).

According to another preferred embodiment, the mobile base is a motorized vehicle with 1 or 2 wheels, which has a substantially humanoid shape, namely equipped with a lower part (articulated leg or rigid member) linked to the, or both , wheel, and an upper part (torso) linked to the lower part by a so-called hip joint; at the top of the upper part (torso) is the shoulder joint, where the arm, or both, is fixed, itself consisting of 2 to 5 segments (110) and at least one gripper (120 ); at least one of the motors of the robot (3) is equipped either with an internal brake, integrated into the motor or arranged a posteriori, or with a dedicated mechanism (190) for blocking the degree of freedom authorized by said motor;

The accompanying drawings illustrate the invention:

represents a general view of the robot, according to the device of the invention.

represents a preferred embodiment of the device corresponding to a manipulator robot for example for logistics.

represents a preferred embodiment of the device corresponding to a manipulator robot, for example for logistics, with two complementary arms (1).

represents a preferred embodiment of the device corresponding to a robot of the humanoid type on wheel(s). According to a non-illustrated embodiment, this robot can have one or more articulated legs as its lower part, or simply crutches.

represents a logical diagram of information flows and the main processing phases.

There illustrates the simplified diagram of the operation of the algorithms (320), which are at the heart of the organization of the control software (300).

There is a detailed, but not exhaustive, overview of the general and logical organization of the algorithms (320).

There represents the robot (3) equipped with an arm (1) and a mobile base (2). The articulation (280) is composed of a motor (250), positioned vertically and which can be integrated into the base, controlling the rotation along the vertical axis (Oz) of the arm (1). If necessary for more clarity in the management of the indices, it will be possible to give the motor (250) the index (i=1), because it is the first motor to control the arm (1). The orientation of the first segment (110_1) is therefore controlled by the motor (250_1); this first segment (110_1) being made up of a short horizontal segment encompassing the motor (150_2) which controls the inclination of the segment (110_2). The angle formed between the vertical and the segment (110_2) in the configuration (j) is named (alpha_j_2). The mobility of the arm (1) once the object (4) is carried, also depends on the segment (110_3) controlled by the motor (150_3), which forms an angle (alpha_j_3) with the vertical. Motors (150_2) and (150_3) retain their driving capacity if the torque required for them to rotate is less than their maximum torque (or nominal if the frequency of use is very high). It is the algorithms (320) that manage, among other things, this angle (alpha_j_i).

In the figure the joints (180_4) and (180_5) of the arm are locked, typically by a dedicated locking mechanism (190). In this way, the arm has a large working area, while being able to handle heavy loads via the motors (150_2) and (150_3).

In addition, the environment (6) presents an obstacle, which forces the arm to grasp the object (4) horizontally; this obstacle is typically the door of an automatic machine: CNC or injection machine. Finally, the mobility of the robot (3) is controlled by the motors (250) of the wheels of the base (2), i.e. even without movement of the arm (1), the robot is able to move the object (4) door. Alternatively, the algorithms (320) can generate an intermediate target, and place the object (4), once past the environmental obstacle (6) (namely the aforementioned door) on a nearby table, or on the platform (7) dedicated to the robot (3). The robot (3) can then seize the object (4) in a new configuration either suitable for transport, or to reach the next target...

There represents the robot (3) provided with an arm (1) and a mobile base (2), this is a preferential mode of the device for parts transport uses, known as logistics. In this variant, the base is rectangular in shape, with a raised part, which supports the arm (1), and a lower part, called a platform (7) which is characterized by a fork shape, capable of moving according to the vertical axis. The transport container (710) is represented by a pallet in European format, and is supported by the fork. Opposite the arm (1) is located on the base (2) a counterweight, which balances the masses, in the case where the object (4) carried would be heavy.

The arm (1) consists of 4 long segments (110) each driven by a motor (150), a short segment (110) driven by the motor (250), and a gripper (120), in the form two opposable half-clamps, each moving a motor (150).

The electronic device (310) is shown as being integrated into the base (2). In general, the control software (300) and the electronic device (310) are partially embedded, and partially located on a remote fixed computer, forming part of the local computer network.

There represents the robot (3) provided with two arms (1) and a movable base (2), this is a preferred mode of the device for interior transport uses of heavy or bulky parts. In this illustration, quite similar to [Fig.2], the presence of two arms (1) makes it possible to stabilize the robot (3) during the rotation of the first arm (1), carrying the heavy object (4) , with the second arm (1) resting on the ground. Said rotation of the object (4) is typically a target generated by the algorithms (320) corresponding to a transition between the picking up of an object (4) and its transport on the platform of the robot (3). The transition phase generates forces which could overturn the robot (3) in the absence of an appropriate configuration of the second arm. Alternatively, the robot can grab heavy or bulky objects with its two arms.

There represents a preferred embodiment of the device corresponding to a robot of the humanoid type whose contact with the ground is made by two laterally opposite wheels on the base. A single arm is illustrated, however an equally preferred embodiment would include a second arm, thus making it possible to take objects from either side, or allowing the support of an arm on the ground, or any other element of the environment, in order to stabilize the robot and/or increase its handling or mobility performance.

There illustrates the information flows, and the different phases of information processing leading, as far as possible, to the execution of the assigned action.

There illustrates the simplified diagram of the operation of the algorithms (320), the tasks carried out, as well as the logical tests carried out. The data (333) is the collection of all the information from all the sources available to the robot (3), including the reprocessed, merged information, and the results of the algorithms (320) during the present iteration, and previous ones.

To maintain a good level of understanding of the , the logic gates and the conditions for transmitting information from one algorithm (320) to another are not shown. These elements are described in the claims. The result of the calculations of the algorithms (320) is either the repetition of a calculation loop by the algorithms (320), or the validation of the result obtained for execution of the action assigned by the control software (300). The possible failure of the feasibility phase is not shown in the illustration.

There illustrates the general and logical organization of the algorithms (320), including the transition logic gates between algorithms, resulting either in the execution of the assigned action in an optimal manner, or in the conclusion (according to a certain probability in relation especially with the validation threshold values used) that the robot is not able to perform the assigned action.

Insofar as the algorithm (327) is dependent on the configuration studied. Its functioning is underlying and permanent, and its action is conditional. And considering the difficulty of illustrating the logical relations in a clear and unambiguous way, it seemed simpler that the said algorithm (327) does not appear on this .

Claims (15)

Device composed of a robot (3) and at least one control software (300) implemented by at least one electronic device (310), allowing said robot (3) to perform at least one action, assigned typically by a human operator or by supervision software; characterized in that:

1. the robot (3) consists of at least one arm (1) and a mobile base (2),
2. the control software(s) (300) evaluates the feasibility of the assigned action, and defines the succession of configurations of the robot (3) allowing the assigned action to be executed in an optimal manner,
3. the control software(s) (300) implements a single configuration library (301), consisting of:
   1. only so-called possible configurations, i.e. which respect the mechanical and motor limitations, in static and dynamic, intrinsic to the robot (3),
   2. initially of a quasi-exhaustive collection of so-called relevant configurations, namely which use the motor and manipulative capacities of the robot (3) to their maximum,
   3. of possible configurations resulting from the aforementioned configurations, and elaborated by the control software (300) with a view to the execution of the assigned action.

Device according to [Claim 1] characterized in that:

1. the arm (1) consisting of at least one mobile segment (110), each segment (110) being moved by at least one motor (150); the segments (110) being interconnected by a hinge (180); the arm (1) being provided with at least one sensor (140); the arm (1) being fixed by one of its ends to the base (2) by a hinge (280); the arm (1) being able to integrate at the other end at least one gripper (120) being able to be equipped with at least one motor (150);
2. the base (2) being provided with at least one sensor (240) and at least one motor (250); the base (2) being mobile and provided with at least one contact member (220) with the ground, typically: wheel, caterpillar, articulated leg, crutch or a combination of these elements;
3. at least one articulation (180) (280) being equipped with a brake integrated into the motor (150) (250), or with a dedicated mechanism (190) for blocking the degree of freedom authorized by the said motor of the said articulation , this dedicated mechanism (190) including at least one motor (150) or (250) fixed locally or remotely;
4. the electronic device(s) (310) controlling, directly or indirectly, all of the motors (150) and (250) of the robot (3) in a coordinated manner; the electronic device(s) (310) being equipped with at least one memory (311) for storing information, and with at least one means of communication (312) with one or more local computer networks, and at least one piece of equipment for local interaction with human operators, called the man-machine interface (313).

Device according to at least one of the preceding claims, characterized in that the control software (300) includes and implements algorithms (320), which operate in parallel, and which iteratively use as input data, among other results generated by the algorithms (320) during the present iteration or previous iterations; the algorithms (320) provide the control software (300) with a set of information necessary and sufficient for the optimal execution of the assigned action; the aforementioned set of information is obtained by performing all or part of the following calculations, for each of the targets of the assigned action:

1. synthesize a substantially comprehensive set of constraints relating to the execution of each target
2. develop new possible configurations from the previously evaluated configurations, and adapt them to the set of execution constraints obtained in (i),
3. possibly generate one or more sub-objectives to be integrated into the assigned action, in order to be able to execute it, or in order to optimize its execution,
4. possibly identify in the environment (6) the element or elements on which the robot (3) can exert a force,
5. evaluate the capacity of the robot (3) in each configuration considered, taken from the library (301), to satisfy all the constraints mentioned in (i), and calculate a score representative of this level of satisfaction,
6. compare and classify the configurations satisfying all the execution constraints obtained in (i), for each target, then define, if it exists, the sequence of optimal configurations allowing to execute the complete assigned action, and finally validate, or not, this sequence.

Device according to at least one of the preceding claims, in a preferred embodiment, characterized in that the mobile base has a substantially rectangular shape, is equipped with 3 or 4 wheels, and where one, or two, arms are fixed relatively close on one side of the rectangle, and where:

1. a platform (7) is arranged on the base (2), characterized in that it:
   1. is substantially horizontal
   2. makes it possible to carry a transport container (710), the arm (1) being able to deposit at least one object (4) on said platform, or on said transport container (710),
   3. has a shape adapted to the transport container (710), namely typically a fork substantially identical to that of a conventional pallet truck,
   4. is optionally movable along a vertical axis, thus making it possible in particular to lift from the ground, respectively deposit, said transport container (710),
2. the arm(s) consisting of 2 to 5 segments (110) and at least one gripper (120),
3. a counterweight (290), removable or fixed, is possibly present, and is typically fixed on the side of the base (2) opposite the arm (1).

Device according to at least one of claims [1], [2] or [3], in a preferred embodiment, characterized in that the mobile base is a motorized vehicle with 1 or 2 wheels, which has a substantially humanoid shape, namely provided with a lower part connected to the wheel, or both, and with an upper part connected to the lower part by a so-called hip joint; at the top of the upper part is the shoulder joint, where the arm, or both, is fixed, itself consisting of 2 to 5 segments (110) and at least one gripper (120); at least one of the motors of the robot (3) is equipped with an internal brake, integrated into the motor or arranged a posteriori, or with a dedicated mechanism (190) for blocking the degree of freedom authorized by said motor. Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is an algorithm (322) which, for each target individually, evaluates all the configurations of the library (301), then selects those which satisfy the conditions below:

1. if the robot (3) is able to reach the target respecting all the constraints described in the data (333) associated with said target, then the configuration is said to be plausible,
2. if the sum of the forces external to the robot (3), namely the forces and the moments in the 3 dimensions, in static or quasi-static, resulting from the execution of the assigned action, is almost zero, or alternatively if the resulting acceleration remains under the control of the robot (3), i.e. can be canceled without falling or overturning, then the configuration is said to be balanced,
3. if the robot (3) is able to generate efforts greater than the efforts necessary to perform the assigned action, then the configuration is said to be efficient,
4. if the robot (3) is able to generate forces greater than the forces required for the two transitions below, namely firstly from a rest configuration, or from the configuration of the previous target (n-1) if it exists, to the configuration of the target (n) currently considered, and secondly from the configuration of the target (n) to the rest configuration, or to that of the next target (n+1) if it exists; then the configuration is said to be compatible,

the expression in mathematical form of all of these conditions constitutes the “synthetic constraints” of the robot (3) for the target considered; and the synthetic constraints are stored in the data (333); the configurations satisfying all the synthetic constraints are said to be satisfactory. Device according to [Claim 6], characterized in that a variant of the aforementioned algorithm (322) considers that one or more of the aforementioned conditions is validated a priori, then exempting the control software (300) from check it. Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is one or more algorithms (323) which on the one hand notes individually, for each target, all the configurations evaluated by the or the algorithm (322), and on the other hand assigns an overall score for each possible succession of satisfactory configurations allowing the complete action to be carried out; the notes are stored in memory (311);
the ratings are based on one or more criteria to be optimized, weighted and/or statistically restated; this, or these, criterion, and statistical weighting or reprocessing rule being supplied to the robot (3) and/or generated by the control software (300);
then the algorithm (323) compares the scores on the present calculation cycle of the algorithms (320) with those obtained previously and stored in memory (311), then for each target the configuration having the best score of all the cycles is the configuration said to be optimal for said target, and for the complete action the succession having the best score of all the cycles is the so-called optimal succession,
and finally before launching the optimal succession allowing the execution of the action, the algorithm (323) will have to confirm the validity of the optimal succession by using one or more validation criteria. Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is an algorithm (324) generating for each target, from configurations (j), not necessarily satisfying the synthetic constraints, of the library (301) potentially improved variants (j'); these configurations (j') are stored in the library (301); the algorithm(s) (324) uses at least one of the following five configuration improvement methods:

1. bring the base (2) closer to the target,
2. block one or more articulation using either the brake integrated into the motor (150) (250), or the dedicated mechanism (190), or both,
3. using at least one arm (1) to bring it into contact with an element of the environment (6) and to exert a force on said element;
4. maintain for each motor, the torque necessary for the execution of the assigned action lower than the maximum torque admissible by said motor; this method is applicable for any articulation (180) (280) of index (i) having a degree of freedom in rotation, driven by a motor (150) (250) called (M_i); this method which consists in limiting the lever arm exerted by the resultant of the external forces on the axis of the engine by controlling the angle of the segment (110_i), can to a certain extent transform a configuration (j) not initially satisfactory, into an efficient configuration (j');
   1. in the particular case, which is nevertheless very frequent, where the forces of gravity are preponderant in relation to the other external forces, this method consists in limiting the angle "alpha_j_i" below the motor's angular traction threshold (M_i), according to the equation: (alpha_j_i) <= (sin-1)(Cmax_i/Cres90_j_i)
      this method can be implemented typically by successively adapting the angles (alpha_j_i) of each segment (110_i) starting with that linked to the joint (280);
5. smoothing or equalizing the criterion or criteria to be optimized for each component of the robot (3) by adapting the configuration of the robot (3).

Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is an algorithm (327), which studies each configuration of the robot (3) being calculated for each target, and is activated in the case where at least one arm (1) is available to exert a force on an element of the environment (6);
then said algorithm (327) identifies in the environment (6) the element or elements on which the arm (1) can exert a force, in view of increasing the performance or the stability of the robot (3); for each new element available in the environment (6) at least one new configuration will be generated by the algorithm (324), this or these new configuration (j') will be stored in the library (301). Device according to at least one of the preceding claims, characterized in that the algorithm (324) preferentially generates, for each target individually, at least one configuration (j'), from one or more configurations (j ) from the library (301) selected according to one or more criteria relating to the score assigned by the algorithm (323) during a previous iteration of the algorithms (320). Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is an algorithm (325) making it possible to add at least one new target (n'), called intermediate, after the execution of the target (n) and before the target (n+1), this addition takes place in the following cases:

1. when the algorithm (322) has identified for each target, at least one configuration satisfying all the conditions except "compatibility", then the addition of a dedicated intermediate target (n') allows the transition of the configuration (j) of the target (n), incompatible with the following target (n+1), towards an intermediate configuration (j') for the target (n'), compatible with the configuration necessary for the following target (n+ 1),
2. when among the algorithms (320) there is an algorithm (326), indicating that the transition from the target (n) to the next one (n+1) induces a significant modification of the synthetic constraints compared to those of the previous target (n ) and successive (n+1), then the addition of a target (n') allows an adaptation of the configuration to these different constraints during the transition.

Device according to at least one of the preceding claims, characterized in that among the algorithms (320) there is an algorithm (321) verifying that there exists for each target at least one satisfactory configuration, thus allowing by at least a succession of satisfactory configurations perform the complete assigned action;
and as long as there does not exist for each target at least one satisfactory configuration, then the algorithms (320) reiterate their calculation loop from a configuration library (301) completed, and from a set of targets possibly bigger ;
the algorithm (321) will also implement a counter of the number of iterations performed by the algorithms (320): if this number of iterations reaches a maximum iteration threshold, typically provided by a human operator, then the feasibility of the assigned action will be considered negative. Device according to at least one of the preceding claims, characterized in that the control software(s) (300) implements the set of algorithms (320) firstly as soon as the action is assigned, in order to determine an optimal embodiment according to the data (333) known at the start, and secondarily permanently, during the execution of the action, in order to adapt its execution to the variations inherent to the robot (3) and to its environment, so as to ensure its execution, and to optimize its execution. Device according to at least one of the preceding claims, characterized in that all or part of the algorithms (320), of the control software(s) (300), of the configuration library (301) and of the data (333) are remote , namely embedded on a fixed and remote computer, typically a high-powered server, and which communicates via the local computer network with the robot (3).