Closed rockstarsir closed 1 year ago
Thanns for your interestes.
The inputs are images and point clouds. The outputs are planned trajectories and control signals. You could check the run_step function of our team agent for details.
We predict both to combine the advantages of two output formats. Please check our previous work TCP for detials. The brake is conducted implicitly by end-to-end learning. Of course, people could set additional emergency braker for better safety.
Hi Team,
I am starting my research in the autonomous driving field, could you help me with the following questions about your architecture.
Thanks in advance