Abstract:
Recently, on a road in Henan Province,
an unmanned express delivery vehicle was driving normally, and two men were "hitchhiking" lying on the roof of the vehicle. The video sparked heated discussions among netizens.
The footage showed that this unmanned express delivery vehicle was driving on the road. Two men were lying on the roof of the vehicle. As the vehicle moved forward slowly, social vehicles passed by from time to time. The scene looked quite thrilling. A passerby filmed this scene and posted it online, attracting attention.
In response, Neolithic’s unmanned vehicle customer service staff responded,
If the person bypasses the camera and is not overweight, the unmanned express vehicle system cannot recognize that there is someone on the roof, and they will provide feedback on the relevant situation.
The response from customer service exposed the blind spots of the current L4 unmanned delivery vehicles at the perception level. The on-board camera mainly faces the front and surrounding road conditions to identify the surrounding road conditions. The roof area is a blind spot for perception. The two men happened to "bypass" the coverage of the camera, and the vehicle continued to drive along the established route, completely unaware that it was forced to carry two passengers.
Public information shows that Neolithic unmanned vehicles are currently the leading company in the field of domestic L4 unmanned delivery. They adopt a "vision-based, lidar-assisted" perception technology architecture. Its X3, X6 and other models have been put into operation on public roads in many cities, with a maximum speed of 50 to 60 kilometers per hour, and a single load of express delivery ranging from 400 to 1,000 pieces.
According to technical design, this type of unmanned vehicle can independently complete path planning, obstacle avoidance, traffic light recognition and other operations, but the perception system obviously does not have redundant designs for extreme scenarios such as "people lying on the roof".
From a technical point of view, the perception blind spots of unmanned vehicles are not unsolvable. They can be covered by adding roof-oriented sensors or weight detection devices. However, whether manufacturers have the motivation to increase hardware costs for this is another matter.
Comments