Chongqing- A significant scientific research result, a fusion perception model, based on vision and lidar, was released recently by the joint research center between Chinese e-commerce company JD.com and the Chongqing University of Posts and Telecommunications (CQUPT).
This technology is being actively adapted to the rover environment so that the rover on Mars can perceive and understand the surrounding environment in the future.
This joint research center was officially inaugurated in November to accelerate the construction of the JD Explore Academy Supercomputing Center and cutting-edge technology research and promote the innovation of scientific and technological achievements in the Chongqing Economic and Technological Development Zone.
Combining the industry advantages of JD.com and the talent of the CQUPT to accelerate the research of cutting-edge technologies and promote scientific and technological innovation, the center is now carrying out R&D of critical technologies of the future, such as Mars unmanned rovers and land-air freight logistics vehicles.
According to the relevant person at JD Explore Academy, Martian weather challenges severely challenge visual sensor imaging, including light, temperature, sand, and other factors that result in poor image quality. A single algorithm based on pure visual perception or a lidar algorithm cannot meet the perception task in the Martian environment.
In order to ensure the safe driving of the rover and the timely and accurate completion of the task, the research team proposed a perception model based on a fusion of vision and lidar.
On the one hand, the model fully exploited the unique information of each mode. On the other hand, vision-lidar fusion modes were integrated to complement the two functions. It is believed that this result will boost the development of Mars exploration.
In addition, the joint research center conducted technical research and development on the issues of high cost and insufficient data associated with Mars' environmental data annotation.
Relevant papers have been submitted to the Computer Vision and Pattern Recognition Conference, the top international conference in the field of computer vision, and the key technologies have also been patented.