Unified Prediction and Planning via Conflict-Aware Disjoint Parameter Training

2026-07-22

AI Summary

This work addresses the performance degradation of unified models on resource-constrained edge devices, where conflicting objectives between motion prediction and safety-aware planning lead to suboptimal behavior. To resolve this “skill conflict” in multi-task joint modeling—a problem first identified in this study—we propose Disjoint Parameter Training (DPT), a novel framework that decouples task-critical parameter regions within a shared encoder and integrates them via a sparse merging strategy. DPT enables efficient parameter sharing while preserving task-specific functionality, and it is compatible with various model merging paradigms. Experiments on the JRDB and JTA pedestrian navigation benchmarks demonstrate that DPT significantly outperforms existing approaches, achieving superior multi-task coordination, enhanced safety guarantees, and improved resource efficiency.

Abstract

Accurate motion prediction of surrounding agents and safe motion planning are two closely coupled key tasks for social robot navigation in crowded environments. Deploying these systems on resource-constrained edge devices necessitates compact, unified models that can perform both tasks simultaneously. However, within these compact shared encoders, recent unified models often overlook severe representational conflicts that arise from the distinct objectives of predicting neighbor behaviors versus ego-centric safety planning. To address this issue, we first identify the Skill Conflict$\unicode{x2014}$a phenomenon where overlapping parameter assignments cause distinct tasks to compete for the same weights, preventing the model from fully specializing in individual skills. To resolve this, we propose a novel model-merging-based framework, Disjoint Parameter Training (DPT). DPT mitigates performance degradation caused by Skill Conflict through distributed parameter learning, which separates the key parameter regions of each task while preserving their core capabilities prior to merging. In addition, we observe that sparse merging, which selectively integrates only the most influential parameters for each task rather than combining all task-specific parameters, yields optimal performance by preventing interference among adjacent features and concentrating representational capacity. DPT can be applied in parallel with a variety of merging methods. Evaluated on standard crowd navigation benchmarks (JRDB and JTA), our framework demonstrates superior performance, validating its versatility and effectiveness for safe, resource-efficient robot navigation.

Problem

• motion prediction
• motion planning
• representational conflict
• unified model
• edge devices

Innovation

• Skill Conflict
• Disjoint Parameter Training
• Unified Prediction and Planning
• Sparse Merging
• Model Merging