Friday, March 15, 2024

robotixx talk

=> interactive learning

adaptive planner parameter learning (APPL)

use behavior cloning to tune any navigation stack

context problem, 

parameter learning

parameter policy

motion policy



=> in situ learning

reference trajectory versus real one
learning inverse kinodynamics

off-road autonomous driving

=> reflective learning (learning from hallucination)

self-supervised learning

=> SCAND; socially compliant navigation demonstration, large scale data set. 



No comments:

Post a Comment