Agile Mobility with Rapid Online Adaptation via Meta-learning and Uncertainty-aware MPPI
Dvij Kalaria Haoru Xue Wenli Xiao Tony Tao Guanya Shi John M. Dolan
Carnegie Mellon University
ICRA 2025
Embedded Files
Abstract
Modern non-linear model-based controllers require an accurate physics model and model parameters to be able to control mobile robots at their limits. Also, due to surface slipping at high speeds, the friction parameters may continually change (like tire degradation in autonomous racing), and the controller may need to adapt rapidly. Many works derive a task-specific robot model with a parameter adaptation scheme that works well for the task but requires a lot of effort and tuning for each platform and task. In this work, we design a full model-learning-based controller based on meta pre-training that can very quickly adapt using few-shot dynamics data to any wheel-based robot with any model parameters, while also reasoning about model uncertainty. We demonstrate our results in small-scale numeric simulation, the large-scale Unity simulator, and on a medium-scale hardware platform with a wide range of settings. We show that our results are comparable to domain-specific well-engineered controllers, and have excellent generalization performance across all scenarios.
Numeric sim results
Trajectory traced for 1 selected run
Stats for 10 runs
Lateral errors for 10 runs :-
a: Without adaptation
b: Without adaptation but correct params incorrect delay
c: With adaptation random initialization
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
Trajectory traced for 1 selected run :-
a: Without adaptation
b: Without adaptation but correct params incorrect delay
c: With adaptation random initialization
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
Unity results
Lateral error/lap time
Lateral error/lap time
Steer bias environment
GoKart environment
Mud terrain environment
Banked-oval racetrack environment
RC car results
Trajectories followed :-
With steer bias:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
Plastic tires:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization (Crashed at t=50.3s)
With towed weights:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
With mat surface:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
With taped tires:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
With normal surface:-
d: With adaptation + MAML initialization + without uncertainty estimation
e: With adaptation + K=0 based initialization + with uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
Some videos :-
With towed weights:-
Plastic tires:-
d: With adaptation + MAML initialization + without uncertainty estimation
f: With adaptation + MAML initialization + with uncertainty estimation (Ours)
g: APACRace initialization
Page updated
Google Sites
Report abuse