Destination support restoration improves robot human path predictions
Destination Support Restoration for Finite-Set Multimodal Trajectory Prediction
RoboticsArtificial Intelligence
Summary
Robots predicting where people will go often choose only a few main paths, missing other possible routes. The authors came up with a way called Destination Support Restoration (DSR) that fixes this by making sure less obvious paths still show up in the predictions. This method adjusts predictions without needing to rebuild the robot’s model or add more guesses, helping robots make better decisions around people. Tests show that DSR improves accuracy in predicting human movement across multiple existing methods.
What this means in practice
- •For robotic navigation teams: Improve pedestrian path predictions by maintaining diverse possible destinations without increasing computational load in robot systems.
- •For autonomous vehicle software developers: Enhance multi-path human trajectory predictions within fixed-size candidate sets to support safer vehicle decision making.
Authors
Fengrui Liu, Jiajun Peng, Duo Peng, Feng Liu
Abstract
Robots operating around pedestrians often reason over a finite set of predicted human futures. Repeated online updates can concentrate this limited prediction budget on dominant destinations and leave plausible alternatives underrepresented or absent, removing those alternatives from the finite representation available to downstream decision making. We introduce Destination Support Restoration (DSR), a causal post-selection operator that repairs destination support without retraining the host predictor or increasing the maintained set size. At a repair step, DSR evaluates a temporary destination-stratified candidate bank from the observed prefix, converts candidate evidence into integer target counts, protects representatives of active modes, and reallocates redundant surplus hypotheses to deficient modes. The maintained and returned sets retain exactly $N$ hypotheses, and DSR replaces at most $\lceilρN\rceil$ entries. Protected representatives preserve current categorical support; lineage-aware particle filters also preserve surviving resampling ancestors. Each replacement reduces the allocation mismatch to the evidence-driven target by one. On the complete 3,719-trajectory Edinburgh protocol over three seeds, DSR reduces MIF weighted ADE and FDE by 13.36% and 13.30% at $N=64$. Paired integrations with CLiFF, PPT, causal GDTS, Social Informer, and PECNet improve both metrics in every evaluated pair. These results show that finite-set support allocation is a useful prediction-side control point when a fixed hypothesis set serves as the interface to downstream systems.