March | 2021
Figure 1 – Types of sensors used for perception during autonomous driving
Thorough testing of the AV is very critical to avoid any misdetection / miscalculations that can result in life threatening consequences. On-road physical testing is time consuming, costly and requires lot of efforts. After driving millions of miles physically, a vehicle may not encounter some of the scenarios and there are less chances to repeat scenarios. In addition, several factors such as temperature, rain, fog, sunlight, etc. are not in the user’s control. In this case, ‘simulation based virtual test’ is the most feasible and essential during the development of the autonomous vehicle. Once the AV stack achieves a certain level of performance maturity in the simulator, it can be tested later with the vehicle on-road for a limited set of cases to reduce overall time to market. Therefore, Wipro is developing ‘SDV in a box’ - a global scale autonomous vehicle driving simulator used for testing and validating the perception and navigation algorithms of autonomous vehicles. It acts as a testing ground for autonomous vehicles before rolling it out on the road by simulating real time traffic scenarios.
AV simulator testing can only be effective if it is able to mimic the sensor behavior similar to the real world. The sensor models used to test an AV must accurately produce the data/signals generated by real sensors in real situations. Therefore, high fidelity virtual sensor models should be included along with realistic environment to capture the complexity accurately during testing and validation of perception, planning and control algorithms. Based on the level of fidelity requirement, virtual sensors can be modeled using one of the models shown in Table 1.
Table 1 – Virtual sensor model categories for AV simulators
The ‘SDV in a box’ has used some of the above approach to model their virtual sensors which can be used for simulation based validation of the Autonomous vehicle stack. Figure 2 shows the GUI of SDV in a box simulator. On the right side of the simulator, users can observe different visualization options for various sensors incorporated. Figure 2 shows some of the features of sensor models used in Wipro’s simulator. All of these sensor models are user configurable. Moreover, these sensor models also output data in different formats for further use in building AI/ML models to drive the vehicle autonomously. To overcome the limitation of one sensor over the other, a sensor fusion approach is used by combining data from various sensors. For example, to avoid false targets detection (due to sensor misreads/uncertainty), different onboard sensors with overlapping features are fused together using different fusion techniques to provide accurate output, enabling driving automation systems to be more efficient. Therefore, Wipro’s simulator not only allows sensor parameter configuration but also allows positional configuration. This ensures that there is no blind spot surrounding the vehicle.
Figure 2 – Sensors implemented in SDV in a Box AV Simulator
Some of the real time complexities associated with different sensors are:
Sensor models in AV simulators should also mimic the limitations of sensors and effect of weather conditions on sensors. Some of the challenges of different sensors are listed in Table 2.
Table 2 – Limitations of various sensors
SDV in a box offers users the options to take into account weather and time of the day effect in simulation. Weather effect includes options such as clear, wet, cloudy, rainy, fog, and snow. Similarly, time of the day includes options such as afternoon, evening or night. Moreover, user can use any combination of these two to make it more realistic and diversified when it comes to autonomous driving scenarios. A few sample examples of combinations are illustrated in Table 3.
Table 3 – Sample visualization of real-time feeds showing combination of weather and time of day (Blue highlighted buttons on right are the combination)
Sensor models in SDV in a box are a combination of the ideal and probabilistic models. Wipro is working towards making them more realistic by adding associated physics and its accurate interaction with environment. There is always a trade-off between accuracy and complexities. Based on the use case, an appropriate virtual sensor model can be chosen for simulation based testing. SDV in a box has various other sensor models as well, i.e. fish eye camera, depth camera, IMU, GNSS, and other ground truth sensors like obstacle detector, collision detector and lane invasion detector.
Dr. Dattatray Parle
Dr. Parle has over 20 years of experience in different industries. His core engineering experience is in CAD/CAM, CAE, SLM & PLM, whereas his emerging technologies experience is centered on Autonomous Vehicles, Industry 4.0, 3D Printing, Virtual Reality, and Augmented Reality. He has published over 40 papers and delivered several technical talks across national and international forums. He is on the board and is a committee member of several industrial and academic bodies. Dr. Parle has worked across the product product lifecycle for industries like automotive, aerospace, oil & gas, biomedical, and nuclear engineering. He can be reached at dattatray.parle@wipro.com.
Yuvika Dev
Yuvika is a Product and Solution Architect of “SDV in a Box". She has over 12 years of software development experience, with close to 10 years in the automotive domain. She has worked on “In Vehicle Infotainment” (Multimedia, Instrument-Cluster, and Bluetooth connectivity) for Automotive OEMs and Tier-1. She is currently a part of Wipro’s Autonomous Vehicle development program – WiPOD, and specializes in building large-scale simulators for simulation-based validation and testing of Autonomous/ADAS systems. She can be reached at yuvika.dev@wipro.com.
Wipro enables a global semiconductor chip manufacturer with an ultra-low-power wireless sensor
Find how Wipro helped a Global Tire manufacturer increase operational efficiency through a robust PLM system
Wipro helps a global medical device manufacturer with an integrated remote patient monitoring system
© 2021 Wipro Limited |
|
© 2021 Wipro Limited |
Pharmaceutical & Life Sciences