WO2022034610A1

WO2022034610A1 - A safety system and method for real time safety gear detection and warning

Info

Publication number: WO2022034610A1
Application number: PCT/IN2021/050760
Authority: WO
Inventors: Shekhar SUMEET; Jakati CHAITANYA
Original assignee: Tvs Motor Company Limited
Priority date: 2020-08-09
Filing date: 2021-08-09
Publication date: 2022-02-17
Also published as: CN116056957A; EP4193328A1

Abstract

The present subject matter relates to safety system (105) for a saddle ride vehicle (100) for real time detection and warning of one or more safety gear, said safety system comprising of an image capturing device (110), an AI based controller (205), a starter relay (210) and one or more alarm (215). The image capturing device captures image, process the captured images and sends it to said AI based controller embedded with an AI based algorithm (220).The AI based algorithm then processes the data and accordingly either activates said starter relay or said alarm. The system for real time helmet detection and warning monitors continuously in real time with a pre-determined delay (T), whether rider is wearing helmet by capturing images via said image capturing device and sending the captured data to said AI based controller (205) until either said starter relay (210) or said alarm (215) is active.

Description

A SAFETY SYSTEM AND METHOD FOR REAL TIME SAFETY GEAR DETECTION AND WARNING

TECHNICAL FIELD

[0001] The present subject matter relates generally to a vehicle. More particularly, the present invention relates to a system and method for real-time safety gear detection and warning.

BACKGROUND

[0002] Typically, more than half of the vehicles on the roads of developing countries are two wheelers. Also, in many low and low-middle income countries saddle type two wheeled vehicles are the main form of motorized transport as two wheelers can provide a less expensive and a more sustainable form of transport. They are often the primary, or most abundant, form of transport in such low- income countries. The major accidents involving a two-wheeler turn fatal due to head injuries. A two wheeler accident or crash may result in head injuries, through either a direct contact with hard objects or as a result of excessive acceleration/deceleration. The best way to prevent such fatal accidents is to wear a helmet. Although, wearing of the helmet is made compulsory by government regulating bodies, yet it is noted that a huge number of incidents of head injury during accidents occur due to lack of sensitivity or carelessness of the rider for not wearing a helmet.

[0003] Further, various researches have shown that wearing an appropriate helmet improves the chances of survival of the rider by more than forty per cent and also helps to avoid injuries to the riders about seventy percent of the times. Furthermore, in general the motorcycle helmet and other safety gear e.g. knee guard, shin pads, safety jacket, etc. are designed to minimize the risks of all kinds of injuries including fatal injuries like head injury. Standards and regulations have been developed to test the effectiveness of helmets and other safety gear in providing protection. However, the protection can only be available in the first place, only if the rider is wearing one or more safety gear including critical safety gear like a helmet. Thus, there is requirement of a system which actively prompts the rider to wear one or more safety gear including a helmet and ensures in a failsafe manner that the rider is wearing the safety gear.

BRIEF DESCRIPTION OF THE DRAWINGS

[0004] The detailed description is described with reference to the accompanying figures. The same numbers are used throughout the drawings to reference like features and components.

[0005] Figure 1 illustrates a perspective view of a vehicle with a system for real time helmet detection and warning as per the present subject matter.

[0006] Figure 2 illustrates a system architecture showing various critical sub-systems of said system for real time helmet detection and warning.

[0007] Figure 3 illustrates a process flow diagram for a method for real time helmet detection and warning system.

[0008] Figure 4 illustrates a block diagram for a method for real time helmet detection and warning system.

DETAILED DESCRIPTION

[0009] Typically, the helmet detection systems that are available depend majorly on specialized helmets and safety gears adapted to include one or more sensors. In the already known solutions, the helmet detection systems are typically based on the placement of one or more sensors in the helmet and other safety gears to detect the presence of the helmet or safety gear on the rider. In general, such systems can control various actions such as sending a warning signal or preventing the vehicle from starting. However, such known arts have inherent lacunae that they are heavily dependent on the availability of a helmet or a safety gear in which said sensors are placed for the detection. Also, these known systems, are not as efficient and even fail to perform required actions when the helmets or safety gears along with said sensors are not present. Many times, it can be a situation when the nder is in a hurry and uses a helmet or safety gear that is not configured to be a part of the detection system, in such scenarios the alternate helmet or safety gear does not have one or more sensors as required and therefore the detection system fails to receive signal from the helmet or the safety gear as required. Thus, there is a need for a system that eliminates the dependency of placing the detection systems on the safety gears. Also, there is a need for a system that eliminates the usage of one or more sensors on the safety gears or otherwise, as required by the earlier known solution.

[00010] The present invention aims to solve the problems as faced by the existing systems. One of the objectives of the present invention is to reduce motorcycle casualties by encouraging the use of safety gear e.g. a helmet. The present invention is configured with a facial recognition based vehicle immobilizer and a rider monitoring system towards achieving rider safety. Further, as per an aspect of the present invention, the safety system is configured with a vision based system that is intrinsic part of a vehicle and does not depend on additional sensors to be placed on the one or more safety gear to detect the presence of the safety gear and communicate with the vehicle. Further, the proposed system as per the present invention is part of an active safety system which would offer enhanced rider safety. Accordingly, as per an embodiment of the present invention, a safety system and method for real time helmet detection and warning is disclosed. In one of the embodiments of the present invention said system for real time one or more safety gear detection and warning is positioned on a vehicle and is not a part of the safety gear itself. Thus, eliminating completely the need of a specially configured safety gear to send signals to the system. Furthermore, as per an embodiment of the present invention, said system, for real time one or more safety gear detection and warning, does not depend on availability of one or more sensors fitted on the safety gear.

[00011] In one of the embodiments of the present invention, said system for real time safety gear detection and warning includes an Al (artificial intelligence) based controller and analyser, an image capturing device for capturing the image of the nder and pillion, a starter relay and an alarm. As per an embodiment of the present subject matter, said image capturing device transmits the data to said Al based controller. In one of the embodiments of the present invention, said Al based controller runs an Al (artificial intelligence) based algorithm for checking whether the rider and pillion are wearing a one or more safety gear.

[00012] As per an embodiment of the present invention, said Al based algorithm works on principle of deep neural networks. In one of the embodiments of the present invention, said Al based algorithm creates a region of interest known as bounding box and the portion of the image inside the bounding box is extracted. According to an embodiment of the present invention, the extracted image from the Al based algorithm is provided as an input to the Deep Neural Network model. Thus, the Al based algorithm ensures faster image processing and consumes less memory in the Al based controller (processor), thereby making the system for real time safety gear detection and warning efficient. According to an embodiment of the present invention, the extracted image goes through an image classifier which is trained to detect face with and without one or more safety gear. The training is dynamic, i.e. the system for real time safety gear detection and warning, initially learns with a plurality of images as available in the Al based controller and further continuously learns by receiving more data in the real time. Therefore, said system for real time safety gear detection and warning as disclosed in the present invention requires fewer images for initial training and is faster with comparatively lesser training time in relation to the conventional system. As per an embodiment of the present invention, the trained image classifier in the Al based algorithm, classifies the input image as face with helmet or without helmet with a probability score. In one of the embodiments of the present invention, the Al based algorithm based on the probability score makes a decision that whether the image is with helmet or without helmet. As per an embodiment of the present invention the Al based algorithm along with the Deep Neural Network Model forms the Al core of the system for real time safety gear detection and warning. Thus, the Al based controller of the system for real time safety gear detection and warning ensures that the system for real time safety gear detection and warning is adaptable, i.e. it can learn based on the gathered data. As per an embodiment of the present invention the system for real time safety gear detection and warning can boot up and function within 30-50 seconds once the system for real time safety gear detection and warning is activated.

[00013] As per an embodiment of the present invention, on successfully detecting the one or more safety gear on the rider and pillion said Al based controller activates the starter relay. Accordingly, the starter relay enables the engine of the vehicle to start. In one of the embodiments of the present invention, if the Al based algorithm is not successful in detecting the one or more safety gear it will disable the engine starting system via the starter relay and vehicle will fail to start.

[00014] However, a situation may arise when the rider or pillion can remove the one or more safety gear once the vehicle starts, this is not ideal and therefore must be prevented. Therefore, as per an embodiment of the present invention, after the system detects the one or more safety gear and the vehicle starts, the system for real time safety gear detection and warning performs continuous monitoring of the rider at a pre-determined interval of time. In one of the embodiments of the present invention at a certain time period, if the system detects that the rider is not wearing the one or more safety gear, it will switch on a signal along with an audible alarm. According to an embodiment of the present invention, said image capturing device captures image at a certain frame per second. In one of the embodiments of the present invention, the Al based controller is embedded with said Al based algorithm and performs the function as an Al (artificial intelligence) machine. As per an embodiment of the present invention, said starter relay acts as a switch to enable and disable the engine based on the input received from said Al based controller. In one of the embodiments of the present invention, said alarm can be either a sound producing member, a visual signal or both.

[00015] As per an embodiment of the present invention, said image capturing device is positioned in close proximity of an instrument cluster of the vehicle. The placement of said image capturing device in close proximity of an instrument cluster provides a better view of the rider when the rider is in a seated position on the vehicle. In one of the embodiments of the present invention, said Al based controller and the starter relay is positioned along with the conventional wiring harness either inside or in close proximity of the instrument cluster of the vehicle. As per an embodiment of the present invention, the system for real time one or more safety gear detection and warning starts when the ignition key of the vehicle is turned ON or the vehicle is fed with a slight movement. The system for real time one or more safety gear detection and warning activates the image capturing device. The image capturing device then sends data to the Al based controller, the Al based controller processes the data received with the Al based algorithm. In one of the embodiments of the present invention, based on the Al based algorithm output, the Al based controller either activates the starter relay or triggers the alarm for warning. According to an embodiment of the present invention, once the starter relay ignites the power unit of the vehicle, the system for real time one or more safety gear detection and warning performs continuous monitoring of the rider at a pre-determined interval of time with a delay (T). In one of the embodiments of the present invention at a certain time period, if the system detects that the rider is not wearing the one or more safety gear, it will switch on a signal along with an audible alarm. Therefore, the system for real time one or more safety gear detection and warning according to the present invention processes the data in real time and prompts the rider whenever required, without additional sensors and is less expensive. As per an embodiment of the present invention, the system for real time safety gear detection and warning can be configured to store any safety violation data on an one or more of an on-board storage unit and a external storage unit e.g. a portable device or a cloud, wherein the violation can be used to reward good riding practices, used for determining insurance claims, bonuses, fleet management and others.

[00016] The present invention along with all the accompanying embodiments and their other advantages would be described in greater detail in conjunction with the figures in the following paragraphs. The present subject matter is further described with reference to accompanying figures. It should be noted that the description and figures merely illustrate principles of the present subject matter. Various arrangements may be devised that, although not explicitly described or shown herein, encompass the principles of the present subject matter. Moreover, all statements herein reciting principles, aspects, and examples of the present subject matter, as well as specific examples thereof, are intended to encompass equivalents thereof.

[00017] Figure 1 illustrates a perspective view of a vehicle (100) with a system for real time safety gear detection and warning (105) (not shown), as per the present subject matter. Said vehicle (100) illustrated, has a step-through type frame assembly. However, it is be noted that the exemplary vehicle can be any motorized vehicle. In one of the embodiments of the present invention, said vehicle (100) an instrument cluster (115) is provided on a handle bar (120) to display various information and warning signals. As per an embodiment of the present invention, an image capturing device (110) is positioned in close proximity of said instrument cluster (115) provided on said handle bar (120) of said vehicle (100). Thus, the placement of said image capturing device (110) in close proximity of said instrument cluster (115) provides a better view of the rider, when the rider is in a seated position on said vehicle (100). In one of the embodiments of the present invention, an Al (artificial intelligence) based controller (205) (not shown) and a starter relay (210) (not shown) is positioned along with the conventional wiring harness either inside or in close proximity of said instrument cluster (115) of said vehicle (100).

[00018] Figure 2 illustrates a system architecture showing various critical sub-systems of said system for real time one or more safety gear detection and warning (105). As per an embodiment of the present invention, said system for real time one or more safety gear detection and warning (105) includes said image capturing device (110), said Al based controller (205), said starter relay (210) and an alarm (215). In one of the embodiments of the present invention, said image capturing device (110) acts as an environmental sensor and captures the image. According to an embodiment of the present invention said image capturing device (110) then converts the captured image data e.g. a grey scale matrix and sends it to said Al based controller (205). In one of the embodiments of the present invention, said Al based controller (205) feeds the image data as received from said image capturing device (110) to an Al based algorithm (220). As per an embodiment of the present invention, said Al based algorithm (220) then processes the image data and accordingly generates an output. In one of the embodiments of the present invention, said starter relay (210) acts as a driver to toggle a starting system (not shown) of the power unit (not shown) of said vehicle (100). According to an embodiment of the present invention said alarm (215) can be a sound producing member, a visual signal or both.

[00019] As per an embodiment of the present invention, said Al based algorithm (220) works on principle of deep neural networks. In one of the embodiments of the present invention, said Al based algorithm (220) creates a region of interest known as bounding box and the portion of the image data inside the bounding box is extracted. According to an embodiment of the present invention, the extracted image data from said Al based algorithm (220) is processed using the Deep Neural Network model. Thus, as per an embodiment of the present invention, said Al based algorithm (220) ensures faster image processing and consumes less memory in the Al based controller (205) (processor), thereby making the system for real time safety gear detection and warning (105) efficient. According to an embodiment of the present invention, the extracted image goes through an image classifier which is trained to detect face with and without one or more safety gear. As per an embodiment of the present invention the training is dynamic, i.e. said system for real time safety gear detection and warning (105), initially learns with a plurality of images as available in said Al based controller (205) and further continuously learns by receiving more data in the real time. Therefore, said system for real time safety gear detection and warning (105) as disclosed in the present invention requires fewer images for initial training and is faster with comparatively lesser training time in relation to the conventional system. As per an embodiment of the present invention, the trained image classifier in said Al based algorithm (220), classifies the input image data, as rider with one or more safety gear or without one or more safety gear with a probability score. In one of the embodiments of the present invention, said Al based algorithm (220) based on the probability score makes a decision that whether the image data is with one or more safety gear. As per an embodiment of the present invention said Al based algorithm (220) along with the Deep Neural Network Model forms the Al core of said system for real time safety gear detection and warning (105). Thus, said Al based controller (205) of said system for real time safety gear detection and warning (105) ensures that said system for real time safety gear detection and warning (105) is adaptable, i.e. it can learn based on the gathered data in real time. As per an embodiment of the present invention said system for real time safety gear detection and warning (105) can boot up and function within 30-50 seconds when said system for real time safety gear detection and warning (105) is activated.

[00020] Figure 3 illustrates a process flow diagram for a method for real time helmet detection and warning (105) system. First step (301) involves turning the ignition ON. Accordingly, in the next step (302) said instrument cluster (115) is turned ON and said safety system for real time one or more safety gear detection and warning (105) is activated. The next step (303) involves, capturing of images at a pre-determined frame per second speed via said image capturing device (110). The images thus captured by said image capturing device (110) are then converted to a reduced compressed data e.g. grey scale matrix and sent to said Al based controller (205) in step (304). In one of the embodiments of the present invention, the next step involves said Al based controller (205) feeding the image data as received from said image capturing device (110) to said Al based algorithm (220) for helmet detection. The next step (305) is to check whether a helmet is detected via said Al based algorithm (220). As per an embodiment of the present invention, said Al based algorithm (220) then processes the image data and accordingly generates an output. If the one or more safety gear is detected via said Al based algorithm (220), said starter relay (210) is activated in step (306). Said starter relay (210) then turns ON said power unit (not shown) as shown in step (308). Once said power unit is ON, said image capturing device (110) continues to capture images at an interval at predetermined time delay (T) in step (309). The images thus captured by said image capturing device (110) are then converted to a compact data and sent again to said Al based controller (205). The loop is continued with an interval of delay (T) for continuous real time monitoring of helmet detection until the power unit is ON or in running condition (309).

[00021] However, in the step (305) if said Al based algorithm (220) fails to detect a helmet, said Al based algorithm (220) activates said alarm (215) as shown in step (307) which can be a sound producing member, a visual signal or both. The next step is to again send the control command to said image capturing device (110) to capture images and send data to said Al based algorithm (220). This loop is maintained till the ignition of said vehicle (100) is ON. Thus, with the steps as described herein, said system and method for real time one or more safety gear detection and warning (105) continuously detects the presence of one or more safety gear in real time and warns the rider if the safety gear is missing or not detected. Therefore, ensuring the safety of the rider. Figure 4 illustrates a block diagram for said system for real time helmet detection and warning (105). According to an embodiment of the present invention, said system for real time one or more safety gear detection and warning (105) includes said image capturing device (110), said Al based controller (205), said starter relay (210) and said alarm (215). In one of the embodiments of the present invention, said Al based controller (205) is embedded with said Al based algorithm (220) for realtime helmet detection and warning. As per an aspect of the present invention, system for real time one or more safety gear detection and warning (105) can be configured or pre-calibrated with various types of safety gear and said Al based algorithm (220) will continue to learn from different variety of safety gears used by rider. Additionally, as per another aspect, said system for real time one or more safety gear detection and warning (105) can be configured to permit selection of one or more safety gear to be mandatory and optional as per needs of safety e.g. helmet can be configured to be a mandatory safety gear while jacket may be optional, thereby leading to disabling of vehicle being driven if helmet is missing and merely giving an alert indication if jacket is missing as so on.

[00022] Many modifications and variations of the present subject matter are possible in the light of above disclosure. Therefore, within the scope of claims of the present subject matter, the present disclosure may be practiced other than as specifically described.

List of Reference Numerals:

100 Vehicle

105 System for real time safety gear detection and warning

110 Image capturing device 115 Instrument cluster

120 Handle bar

205 Al based controller

210 Starter relay

215 Alarm 220 Al based algorithm

T Delay

Claims

We Claim:

1. A system for real time one or more safety gear detection and warning (105) for a saddle ride vehicle (100), said system for real time one or more safety gear detection and warning (105) comprising: an image capturing device (110); an Al based controller (205); a starter relay (210); and an alarm (215); wherein, said image capturing device (110) captures image and process the captured image data; wherein, said Al based controller (205) is embedded with an Al based algorithm (220); wherein, said Al based controller (205) receives input from said image capturing device (110) and feeds to said Al based algorithm (220); wherein, said Al based algorithm (220) then processes the received data and accordingly generates an output; wherein, said starter relay (210) receives input from said Al based controller (205) and if a one or more safety gear is detected by said Al based algorithm (220), said starter relay (210) activates a starting system of the power unit of said saddle ride vehicle (100); wherein, if the said Al based algorithm (220) does not detect one or more safety gear, said Al based algorithm (220) activates said alarm (215).

2. The invention as claimed in claim 1, wherein, said alarm (215) is one or more of a sound producing member, a visual signal or both.

3. The invention as claimed in claim 1, wherein, said image capturing device (110) converts the captured image to a grey scale matrix.

4. The invention as claimed in claim 1, wherein, said image capturing device (110) is positioned in close proximity to an instrument cluster (115) on a handle bar (120) of said vehicle (100).

5. The invention as claimed in claim 1, wherein, said image capturing device (110) continues to capture images with a pre-determined time delay (T) until said starter relay (210) is active and the image capturing device (110) continues to process the captured images and send it to said Al based controller (205) until the power unit is in running condition.

6. The invention as claimed in claim 1, wherein, said image capturing device (110) continues to capture images with a pre -determined delay (T) until said alarm (215) is active and the image capturing device (110) continues to process the captured images and send it to said Al based controller (205) until the power unit is in running condition.

7. The invention as claimed in claim 1, wherein, said system for real time safety gear detection and waning (105) is configured or pre-calibrated with various types of safety gear and said Al based algorithm (220) continues to learn from different variety of safety gears used by rider.

8. The invention as claimed in claim 1, wherein, said system for real time safety gear detection and waning (105) is configured to select one or more safety gear as mandatory and optional based on pre-determined safety requirements.

9. The invention as claimed in claim 1, wherein, said Al based algorithm (220) works on principle of deep neural networks.

10. A method for real time safety gear detection and waning, said method comprising: turning ignition ON of a vehicle (100) (step 301); activating an instrument cluster (115) and activating an image capturing device (110) for capturing the images (step 302) and converting the captured images to a grey scale matrix (step 303); sending the processed images to an Al based controller (205) and said Al based controller (205); processing said input received by said Al based controller (205) via an Al based algorithm (220) (step 304); sending signal to a starter relay (210) (step 305), if said Al based algorithm (220) detects one or more safety gear and said starter relay (210) starting system of power unit of said vehicle (100) (step 306); sending signal to an alarm (215), if said Al based algorithm (220) fails to detect a one or more safety gear (step 307); monitoring continuously in real time with a pre-determined time delay (T), whether rider is wearing one or more safety gear by capturing images via said image capturing device (110) and sending the captured image to said Al based controller (205) until either said starter relay (210) or said alarm (215) is active;

11. The method as claimed in claim 10, wherein said method includes configuring various types of safety gear in said safety system (105).

12. The method as claimed in claim 10, wherein said method includes, configuring said Al based algorithm (220) to learn continuously from different variety of safety gears used by rider and configuring said safety system (105) to select one or more safety gear as mandatory and optional based on pre-determined safety requirements.

13. The method as claimed in claim 10, wherein said method includes, said Al based algorithm (220) working on the principle of deep neural networks.

15