DE102018211943A1 - Radar system for collision avoidance of a robot - Google Patents

Abstract

The invention relates to a method for avoiding or at least mitigating a collision of a robot with an object using a radar device, the object and at least part of the robot moving relative to one another. The method comprises the following steps: a) emitting electromagnetic waves into at least part of the surroundings of the robot, b) detecting a signal which is reflected or emitted from the object to the antenna due to the impact of the emitted electromagnetic waves on the object, c) generating an image of the object according to the SAR principle by means of the processor and determining the relative distance and speed between the robot and the object, d) determining whether there is a risk of a collision between the robot and the object, and e) if a collision is imminent, changing the planned movement of the robot into such a movement that the collision is avoided or at least mitigated. The invention further relates to a system comprising a robot and a radar device, the system being suitable for performing the above-mentioned method.

Description

The invention relates to a method for avoiding or at least mitigating a collision of a robot with an object, the object and at least part of the robot moving relative to one another. The invention further relates to a system comprising a robot and a radar device, the system being suitable for carrying out such a method.

In a production automated by robots, the challenge has always been to avoid a collision of the robot with objects in the vicinity of the robot. In the context of this patent application, an object is understood to mean any object or living being, in particular also people.

For stationary robots, i.e. Robots that do not move relative to their surroundings traditionally have fences or cages around the robot. The aim of the fences or cages is to prevent an object from penetrating into an area in the immediate vicinity of the robot, in which a collision between the robot and the object could occur. Alternatively, virtual “fences” or security zones are known, which are implemented, for example, by light barriers or sensors. If the object crosses a predetermined, imaginary line in space, this is reported to the robot's safety system, so that appropriate countermeasures, such as acoustic or optical warning signals, the robot's standstill or the robot's active turning away with respect to the approaching object, can be initiated.

Movable robots, i.e. In principle, robots that move relative to their surroundings risk to collide with both stationary and objects that also move relative to their surroundings. Physical fences or cages are unsuitable for moving robots, as this may have to be used to block large areas of a production hall. Such measures are less practical for increasingly densely populated production halls and the increasing interaction between robots or robots and humans.

Similar challenges generally exist for robots, which are fixed to the floor by means of a base, for example, but have moving parts, e.g. Have swivel or gripper arms. Here, too, there is a risk of collisions between the moving parts of the robot and objects in the vicinity of the robot.

As a logical consequence, mobile robots or stationary robots with moving parts nowadays often have devices for setting up and monitoring a virtual security zone of the robot. These devices include e.g. Light or laser barriers. If the distance between the object and the robot falls below a certain minimum distance, this is reported to the robot's safety system, so that appropriate countermeasures can be initiated, as already described in the context of stationary robots. An alternative to the virtual fences are cameras attached to the robot. These can capture the surroundings of the robot and create an environment map based on this. The area map is advantageously updated continuously by the robot's security system.

One disadvantage of light or laser barriers is the relatively high energy that such systems consume continuously. Cameras also need good visibility, which cannot always be guaranteed reliably or only with great effort in a dusty production environment or with insufficient illumination of the production hall. Another limitation of image evaluating, i.e. Camera-based surveillance is the frequency at which the picture is taken. This depends heavily on the design of the camera and the associated processor, but in practice it can be quite significant with high relative movements between the robot and the object.

It is an object of the present invention to overcome the disadvantages of existing systems and methods. In particular, a concept is to be developed of how reliably and inexpensively collisions between a robot and an object, the object and at least part of the robot moving relative to one another, can be avoided or at least mitigated.

This object is achieved by the device and the method of the independent claims. Advantageous further developments and variations of the invention can be found in the subordinate claims.

1. a) elektromagnetische Wellen mittels einer Aussendeeinheit in zumindest einen Teil der Umgebung des Roboters ausgesendet werden,
2. b) ein Signal, das aufgrund des Auftreffens der ausgesendeten elektromagnetischen Wellen auf das Objekt von diesem zu einer Empfangseinheit reflektiert oder ausgesendet wird, detektiert wird,
3. c) ein Abbild des Objekts nach dem SAR-Prinzip mittels des Prozessors erzeugt und die relative Entfernung und Geschwindigkeit zwischen dem Roboter und dem Objekt bestimmt wird,
4. d) ermittelt wird, ob eine Kollision zwischen dem Roboter und dem Objekt droht, und
5. e) falls eine Kollision droht, die geplante Bewegung des Roboters in eine solche Bewegung geändert wird, dass die Kollision vermieden oder zumindest abgemildert wird.

According to the invention there is provided a method in which

1. a) electromagnetic waves are emitted into at least part of the surroundings of the robot by means of an emitting unit,
2. b) a signal is detected which is reflected or emitted from the object to a receiving unit due to the impact of the emitted electromagnetic waves on the object,
3. c) an image of the object according to the SAR principle is generated by the processor and the relative distance and speed between the robot and the object is determined,
4. d) it is determined whether there is a risk of a collision between the robot and the object, and
5. e) if a collision threatens, the planned movement of the robot is changed to such a movement that the collision is avoided or at least mitigated.

The present invention therefore proposes a completely new and unexpected way to avoid collisions between a robot and objects or people moving relative to it: a radar-based generation of an image of the surroundings, with a radar with a synthetic aperture (SAR - "Synthetic Aperture Radar") is used. The high resolution of the generated image of the environment, which can be achieved by means of SAR, is decisive, since a potentially high-resolution, detailed image of the environment, in particular, can also be used to precisely reproduce the objects causing the collision. This allows a reliable assessment of whether a collision between the robot and surrounding objects is to be taken care of and, if so, to initiate appropriate countermeasures.

The electromagnetic waves which are emitted by the radar device of the robot in the context of the method according to the invention are advantageously in the radio frequency range and are also referred to as radio waves or radio waves or Hertzian waves. Due to the available frequency bandwidth, radar systems in the GHz range (Giga-Hertz range) are typical.

The use of a synthetic aperture radar is known from remote sensing. For this purpose, a corresponding radar device is placed on board an airplane, cruise missile or satellite. The beam direction of the antenna of the radar device must be perpendicular to the direction of movement of the aircraft / satellite or the like. his. The principle of the synthetic aperture is to replace the snapshot of a large antenna with many shots of a small, moving antenna. In the course of this movement, every object in the target area is illuminated with a variable angle of view and recorded accordingly. If the path of the real antenna is known with sufficient accuracy and the scenery is immobile, the aperture of a large antenna can be synthesized from the intensity and phase of the received radar echoes, thus achieving a high spatial resolution in the direction of movement of the antenna.

The radio waves can be emitted in a focused or unfocused manner. Only the radio waves reflected on the object can be detected by the antenna of the radar device (primary radar). Alternatively, a transponder can also be located on the object, which responds to the radio waves emitted by the radar device and in turn sends a signal back to the radar device (secondary radar). Pulse or continuous wave radar devices are possible as radiation sources for the radio waves.

On the one hand, it is possible for the radar device to be stationary relative to the surroundings of the robot and for the object to be detected to move relative to the surroundings. The radar device can, for example, be attached to the ceiling of the production hall in which the robot is located. This is particularly useful for stationary robots and robots with a small range of motion, for which the radar device is advantageously placed not far from the robot or near the center of its range of motion.

On the other hand, the robot can also move relative to its surroundings and (relative to the surroundings) detect stationary objects. In this case, the radar device is advantageously attached to the robot so that it moves with the robot.

Furthermore, the robot can also be a robot which is inherently firmly connected to the floor (or the ceiling) and has a part, for example a swivel or gripper arm, which is movable relative to the surroundings of the robot. For example, a person could approach the robot that was turned off to perform some activity near the robot. When the robot is activated, the person could still be within range of the robot's swivel or gripper arm. If the swivel arm or gripper arm now moves in the direction of the person, the radar device would detect the decreasing distance between the swivel arm or gripper arm and the person and would also recognize that if the robot continues to move the arm and the person is assumed to have stopped at the current position Collision between the robot arm and the person is imminent. The robot can be designed in such a way that the planned movement of the swivel or gripper arm of the robot is then changed, so that the collision is avoided or at least mitigated.

Finally, both the robot as a whole and the object can move relative to the environment. Here too, the SAR principle can be applied and the collision probability between the robot and the object can be determined. Only with both stationary robots as well as a stationary object cannot be imaged using SAR (the same applies in principle if the robot and the object move in the same direction at the same speed, as this also leads to a relative speed of zero) ,

In one possible embodiment, the radio waves can only be emitted in a defined direction relative to the robot. However, the radar device advantageously has a mechanism with which the radio waves can be emitted (in succession) in different directions relative to the robot. This allows an image of a larger part of the environment to be generated, so that objects which may cause collisions can be identified with a higher probability. A mechanism which allows electromagnetic waves to be emitted in different directions could, for example, consist of a rotatable and / or pivotable transmitter.

If the robot is a robot that is movable relative to the surroundings and if the above-mentioned mechanism with a rotatable or pivotable transmitter is to be dispensed with, the radar device can advantageously have a first antenna and a second antenna, which are essentially at a 90 ° angle to one another are aligned. The first and second antennas are preferably located approximately in the plane of movement of the robot. In the context of this patent application, the “movement plane” means the plane in which the moving robot moves.

Particularly good measurement results according to the SAR principle can be achieved if the first antenna is not only arranged perpendicularly to the second antenna and the two antennas are not only essentially in the plane of movement, but if the two antennas are also approximately at a 45 ° angle to the direction of movement of the robot (movable relative to the environment). This enables an optimal detection of the signals reflected or emitted by an object for any objects that are located anywhere in the movement plane of the robot.

It can also be advantageous if the method for avoiding a collision of a robot with an object moving relative to the robot comprises a step in which the absolute position of the robot is determined. This can be done by satellite (e.g. via GPS - Global Positioning System) or odometrically, i.e. by calculating the position of the robot based on the wheel revolutions made if the robot is driven by wheels or based on the steps taken.

The odometric position determination becomes all the more inaccurate the further back the calibration point lies on which the wheel revolutions or steps are counted. In principle, the satellite-based position determination always has the same inherent blur, but requires a relatively large amount of energy and is dependent on the reception of the corresponding satellite data, which is more or less good depending on the production location and position within the production hall. The highest accuracy when determining the position can in principle be achieved by a combination of both methods.

Even if knowledge of the absolute position of the robot is not a necessary prerequisite for successfully carrying out the method according to the invention for collision avoidance, in practice it is often advantageous if, in addition to the relative distance and relative speed between the robot and the object, the absolute position and the absolute speed of the object can be determined.

Another promising variant of the invention, which is particularly suitable for solving the problem of the invention to reliably avoid collisions, is the inclusion of machine learning. Here, the robot should record at least part of its surroundings several times during a learning phase and recognize patterns and regularities in the recorded data. These patterns and laws can then be used to determine a familiar environment. The "determination of the familiar environment" can also be called "empty measurement". It serves to distinguish and detect surprisingly appearing objects with higher accuracy and reliability from known and usually present objects.

Alternatively or additionally, machine learning can also be used in a different context to increase the detection accuracy of the radar device: during a learning phase, the processor generates images of objects in the vicinity of the robot according to the SAR principle in a first step. In a second step, these images are compared with actual images of these objects, which a user assigns to the generated images. Using this method, the system is “trained” and can use the information given by the user as to whether or to what extent a generated image was correct in order to improve its detection accuracy.

It is also possible that the knowledge of repetitive movements of the robot or parts thereof is also used for collision avoidance. This is also known as the knowledge of current and probably future pose of the robot.

The last step of the method according to the invention includes changing the planned movement of the robot such that a collision of the robot with the object is avoided or at least mitigated if such a collision threatens due to the positions of the robot and the object and the relative movement between them. The change in motion of the robot can include an “active” turning away of the robot or parts thereof from the object. The change in movement can also include a standstill of the robot and persistence in it until a signal to start movement has been given. The change in movement can also involve the robot moving to a different movement path than originally planned, in order thereby to avoid a collision between the robot and the object. Finally, it should be mentioned that in some cases a collision between the robot and an object cannot be completely avoided, but the consequences of the collision can be mitigated by the method according to the invention and an advantageous change in movement of the robot in the last method step. This is the case, for example, when an object, for example a person, moves so suddenly and close to the robot that the robot still initiates an evasive movement, but a collision or a streak of the robot and the person is not completely prevented can.

• - eine Aussendeeinheit, die elektromagnetische Wellen in zumindest einen Teil der Umgebung des Roboters aussenden kann;
• - eine Empfangseinheit, die ein Signal, das aufgrund des Auftreffens der elektromagnetischen Wellen auf dem Objekt von diesem zur Antenne reflektiert oder ausgesendet wird, detektieren kann; und
• - einen Prozessor, der ebenfalls ein Teil des Radargeräts ist, und der ein Abbild des Objekts nach dem SAR-Prinzip erzeugen und die relative Entfernung und Geschwindigkeit zwischen dem Roboter und dem Objekt bestimmen kann.

The invention also relates to a system comprising a robot and a radar device, the system or the robot being suitable for carrying out one of the aforementioned methods. The radar device should in particular have the following components:

• an emitting unit that can emit electromagnetic waves in at least part of the surroundings of the robot;
• a receiving unit that can detect a signal that is reflected or emitted from the object to the antenna due to the impingement of the electromagnetic waves on the object; and
• - A processor, which is also part of the radar device, and which can generate an image of the object according to the SAR principle and can determine the relative distance and speed between the robot and the object.

The transmitter unit and the receiver unit can be realized by a single component, which in particular comprises one or more antennas. In this case one speaks of a monostatic radar system. Alternatively, the two units can also be separate (e.g. having a transmitting and a receiving antenna), which is then referred to as a bistatic radar system.

It may be advantageous to attach the radar device to an exposed part of the robot. In principle, this allows the radio waves of the radar device to be transmitted and received as freely as possible.

In the following, the invention is described concretely with reference to an illustration, which of course is only an example, i.e. represents a possible embodiment of the invention and is in no way restrictive of the claimed scope of protection of the patent.

The figure shows a stationary robot 10 by means of its base 11 stuck to the ground 41 is connected, so it stands on it. On the base 11 is a pedestal 12 arranged. The stand is aligned vertically and around a vertically oriented axis of rotation 121 in the direction of rotation 122 rotatable. One rotation in the direction of rotation shown 122 as well as against this is advantageous. On the base 11 opposite end of the pedestal 12 is a first swivel arm 13 arranged. The first swivel arm 13 is about a first pivot axis 131 regarding the stand 12 pivotable. The swivel movement is in both swivel directions shown 132 possible. Finally, on the base 12 opposite end of the first swivel arm 13 a second swivel arm 14 arranged. The second swivel arm 14 is about a second pivot axis 141 with respect to the first swivel arm 13 pivotable. The swivel movement is in both swivel directions shown 142 possible. As a result, the schematically drawn and exemplary robot points 10 three axes of rotation or swivel. In reality, so-called six-axis robots are common, which have further swiveling and rotating movements and thus enable almost every movement of the robot arm. The illustration of such a six-axis robot has been omitted for reasons of clarity - especially since this is not essential for the functionality of the present invention.

There is also an object in the illustration 30 to see that in this example there is a person. The person approaches in the direction of movement 31 the robot 10 , especially its robot arm. A radar device 20 that is in the area 40 of the robot 10 , namely directly above the robot 10 on the ceiling 42 creates an image of the person and determines their distance to the robot and the relative speed including their direction of movement. It is recognized that there is a risk of a collision between the person and the robot, and consequently the robot arm is turned away to the rear (to the left in the figure). This can cause a collision between the person and the robot 10 be avoided.

Claims (12)

Method for avoiding or at least mitigating a collision of a robot (10) with an object (30) using a radar device (20), the object (30) and at least part of the robot (10) moving relative to one another, and the radar device has at least one transmission unit, a reception unit and a processor, and wherein the method comprises the steps: - emitting electromagnetic waves into at least part of the environment (40) of the robot (10) by means of the emitting unit, Detection of a signal which is reflected or emitted from the object (30) due to the impact of the emitted electromagnetic waves on the object (30), Generating an image of the object (30) according to the SAR principle by means of the processor and determining the relative distance and speed between the robot (10) and the object (30), - Determine whether a collision between the robot (10) and the object (30) threatens, and - If a collision threatens, changing the planned movement of the robot (10) into such a movement that the collision is avoided or at least mitigated. Procedure according to Claim 1 , wherein the radar device (20) is stationary relative to the environment (40) of the robot (10) during the method. Method according to one of the preceding claims, wherein the radar device (20) is placed separately from the robot (10). Procedure according to Claim 1 , wherein the radar device (20) moves during the method relative to the environment (40) of the robot (10). Procedure according to Claim 4 The radar device (20) is attached to the robot (10) moving relative to the surroundings (40). Method according to one of the preceding claims, wherein the method further comprises determining the absolute position of the robot (10). Method according to one of the preceding claims, wherein the method further comprises a first learning phase, in which the robot (10) repeatedly detects at least a part of the environment (40) of the robot (10) and recognizes patterns and regularities in the recorded data, and these patterns and laws are used in determining a familiar environment. Method according to one of the preceding claims, the method further comprising a second learning phase in which the generated images of the objects (30) are compared with actual images of the same objects (30) in order to improve the quality of the images. Method according to one of the preceding claims, wherein the change in the planned movement of the robot (10) takes place when a predetermined distance between the object (30) and the robot (10) is undershot. System comprising a robot (10) and a radar device (20), the radar device (20) having at least one transmission unit, a receiving unit and a processor, and the robot (10) being designed to collide with an object (30) to avoid or at least mitigate, and wherein - The transmission unit can emit electromagnetic waves in at least part of the environment (40) of the robot (10), the receiving unit can detect a signal that is reflected or emitted by the object (30) due to the impact of the electromagnetic waves on the object (30), - The processor can generate an image of the object (30) according to the SAR principle and can determine the relative distance and speed between the robot (10) and the object (30) if the object (30) and at least part of the robot (10) move relative to each other, - The processor can determine whether a collision between the robot (10) and the object (30) threatens, and - If a collision threatens, the planned movement of the robot (10) can be changed into such a movement that the collision can be avoided or at least mitigated. System according to Claim 10 , wherein the transmitter unit and the receiver unit are realized by the same component, namely an antenna. System according to one of the Claims 10 or 11 , The transmitter unit being located on an exposed part of the robot (10), in particular a swivel arm of the robot.

Images