Unicycle Robot Navigation with Contingency Planning on 100 Randomly Generated Cluttered Environments
100Success RateMPPI
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MPPIcontingency awareness=none2026.03 | 100 | 80 | 0.202 | — | 15.4 | 2.6 | 22 | |
| SCRAMPPIResampling-based rollouts (RBR)=true, HJ reach-avoid value function cost=true2026.03 | 100 | 102 | 0.446 | 100 | 0 | 44.4 | 33.6 | |
| SCRAMPPI (no RBR)Resampling-based rollouts (RBR)=false, HJ reach-avoid value function cost=true2026.03 | 92 | 106 | 0.321 | 100 | 0 | 43.8 | 33.6 | |
| C-MPPI-1000Contingency samples (N)=10002026.03 | 84 | 119 | 0.253 | 98.5 | 0.4 | 126.4 | 1,159.7 | |
| C-MPPI-100Contingency samples (N)=1002026.03 | 44 | 126 | 0.144 | 96.3 | 0.4 | 17.2 | 102.7 |