🤖 AI Summary
This work addresses the limited adaptability of existing learning-based path planning methods, which are typically constrained to shortest-path objectives and struggle to accommodate diverse task criteria. To overcome this, the authors propose FlexPath, a two-stage framework: the first stage leverages imitation learning to extract task-agnostic feasible path priors from visual maps, while the second introduces a differentiable Path Shape Objective (PSO) that efficiently adapts paths to varying task preferences without altering their underlying structure. FlexPath is the first approach to decouple path feasibility from task-specific objectives, enabling a single pre-trained model to generalize across multiple planning criteria through objective-level adjustments while remaining compatible with classical planners. Experiments show that on the TMP dataset, FlexPath reduces search overhead by 14.3% compared to TransPath, yields lower average path costs, successfully generalizes to three unseen domains in a zero-shot setting, and achieves a 96.8% success rate in full obstacle avoidance.
📝 Abstract
Recent learning-based path planners use neural networks to process visual map representations and approximate heuristics for classical search algorithms, yielding near-optimal paths with reduced search effort. However, these methods are tied to the shortest-path objective implicit in their supervision, which limits their flexibility to accommodate alternative criteria. We introduce FlexPath, a two-stage framework that decouples feasibility from preference. In Stage 1, we use imitation learning to acquire a task-independent spatial prior over feasible paths from visual map inputs. In Stage 2, differentiable Path Shape Objectives (PSOs) adapt this prior toward task-specific criteria without relearning path structure, requiring only efficient objective-level adaptation. A single pretrained model can be adapted to multiple objectives. For shortest-path planning, FlexPath reduces search effort on TMP by 14.3% compared to the state-of-the-art TransPath, while also finding lower-cost paths on average and demonstrating strong zero-shot generalization across three unseen domains. For obstacle clearance with minimum clearance distance 2, it achieves 96.8% full obstacle avoidance while maintaining low search cost. The framework further extends to semantic-aware avoidance and waypoint guidance via objective-level adaptation, and remains compatible with classical planners at inference time. Data and code are available at https://github.com/FraunhoferIVI/FlexPath.