TU-TESL

• 22

According to foreign media reports, in order to improve the security of the system, MIT has developed a system called shadowcam. The system senses subtle changes in ground shadows to determine whether vehicles or other objects are approaching behind corners. It is reported that researchers have carried out the test of self driving vehicles around the parking lot. The shadowcam system can use the visual perception technology to detect the change of the ground shadow in front of the vehicle, and classify it to determine whether there are hidden mobile vehicles or pedestrians in front. At the same time, the researchers also used an automated wheelchair in the hospital corridor to conduct tests to avoid pedestrians. Both tests were successful. During the test, the self driving vehicle will actively stop to avoid collision after sensing other vehicles approaching. It is reported that part of the performance index of the shadowcam system on the test vehicle is better than that of the traditional lidar. The lidar can only detect visible objects, while the detection speed of the shadowcam system for hidden objects is about 0.5 seconds earlier than that of the lidar. The researchers say that 0.5 seconds ahead of time for fast driving autopilot will greatly enhance the system's coping ability. At present, in order to ensure the consistency of lighting conditions, the system can only be tested in indoor environment, and the ground shadow can be clearly perceived and analyzed.