Algorithms and Frameworks for Multi-Sensor Collaborative Perception in Autonomous Driving

The recent years have seen work done on algorithms and frameworks of multi-sensor cooperative perception. This is an emphasis meant to improve the environmental perception of the autonomous vehicles. The heterogeneity of the multimodal data can be viewed as one of the major challenges during the multi-sensor fusion process. Another major challenge is in difference in the fusion levels. Also, model computational efficiency is also an important research problem. The paper will begin by giving the nature of popular sensors. It goes on to describe pertinent research areas in multi-sensor fusion in autonomous driving. Advanced paradigms of fusion symbolized by BEV and Transformer are analyzed in detail. New vehicle-infrastructure vehicles co-operative fusion structures are discussed too. Lastly, the paper will conclude by summarizing the existing issues and predicting the future trends. A perceived and decision making and planning is one of the possible development directions in the future.