When Should a Robot Replan? Regret-Guided Update Scheduling in Time-Varying MDPs
Overview
Robots operating in non-stationary environments must adapt their policies as dynamics drift, but limited onboard energy and compute constrain how often full state estimation and re-planning can occur. This work formulates the problem within time-varying Markov decision processes (TVMDPs) with a known bound on transition drift rate, modeling execution via a skip-update scheme where the agent estimates the transition kernel by maximum likelihood at chosen update times and reuses the policy between updates under propagated state estimates. The authors analyze dynamic regret growth during skip intervals and derive an online, regret-guided update rule that adaptively allocates the budget. Evaluation in simulated Mars-rover navigation with time-varying slip dynamics and on a Crazyflie quadrotor in indoor obstacle fields shows adaptive allocation outperforms other budgeted baselines.
Generated from attributed reports · Updated 2 hours ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Environment: simulation
“intervals in terms of the properties of the TVMDP and the skip lengths; the resulting bound answers the opening question via an online, regret-guided update rule that allocates the budget adaptively. We evaluate the rule in a simulated Mars-rover navigation task with time-varying slip dynamics and on a Crazyflie quadrotor in indoor obstacle fields. Adaptive allocation outperforms other budgeted baselines.”
Exact source · revision 1Source owner not reported
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsSignalWhen Should a Robot Replan? Regret-Guided Update Scheduling in Time-Varying MDPs
Robots operating in non-stationary environments must adapt their policies as dynamics drift, but limited onboard energy and compute constrain how often full state estimation and re-planning can occur. This work formulates the problem within time-varying Markov decision processes (TVMDPs) with a known bound on transition drift rate, modeling execution via a skip-update scheme where the agent estimates the transition kernel by maximum likelihood at chosen update times and reuses the policy between updates under propagated state estimates. The authors analyze dynamic regret growth during skip intervals and derive an online, regret-guided update rule that adaptively allocates the budget. Evaluation in simulated Mars-rover navigation with time-varying slip dynamics and on a Crazyflie quadrotor in indoor obstacle fields shows adaptive allocation outperforms other budgeted baselines.
Event coverage history
There is not enough continuous observation data to show a trend.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).