HIRO: Heuristics Informed Robot Online Path Planning Using Pre-computed Deterministic Roadmaps

Xi Huang,Gergely Sóti,Hongyi Zhou,Christoph Ledermann,Björn Hein,Torsten Kröger

2024-10-27

Abstract:With the goal of efficiently computing collision-free robot motion trajectories in dynamically changing environments, we present results of a novel method for Heuristics Informed Robot Online Path Planning (HIRO). Dividing robot environments into static and dynamic elements, we use the static part for initializing a deterministic roadmap, which provides a lower bound of the final path cost as informed heuristics for fast path-finding. These heuristics guide a search tree to explore the roadmap during runtime. The search tree examines the edges using a fuzzy collision checking concerning the dynamic environment. Finally, the heuristics tree exploits knowledge fed back from the fuzzy collision checking module and updates the lower bound for the path cost. As we demonstrate in real-world experiments, the closed-loop formed by these three components significantly accelerates the planning procedure. An additional backtracking step ensures the feasibility of the resulting paths. Experiments in simulation and the real world show that HIRO can find collision-free paths considerably faster than baseline methods with and without prior knowledge of the environment.

Robotics

What problem does this paper attempt to address?

This paper attempts to solve the problem of efficiently computing collision - free robot motion trajectories in a dynamically changing environment. Specifically, the researchers proposed a new method - HIRO (Heuristics Informed Robot Online Path Planning) to quickly find collision - free paths in the case of dynamic environmental changes.

### Main problems and solutions in the paper

#### 1. **Background and challenges**
- **Limitations of traditional methods**: Traditional sampling - based methods such as RRT (Rapidly Exploring Random Tree) and PRM (Probabilistic Roadmap) perform well in static environments, but have a slow response speed in dynamic environments and cannot respond to environmental changes in a timely manner.
- **Requirements of new applications**: With the emergence of new application scenarios such as human - robot collaboration, robots need to be able to respond to human actions in real - time, which requires robots to have the ability to quickly plan paths in dynamic environments.

#### 2. **Core ideas of HIRO**
- **Separation of static and dynamic environments**: HIRO divides the robot's workspace into a static part and a dynamic part. The static part includes fixed objects (such as tables, the robot itself), while the dynamic part contains all potentially moving objects (such as operating objects or users).
- **Pre - computing the deterministic roadmap**: Use the static environment information to generate a deterministic roadmap, which can provide a lower - bound estimate of the final path cost as the basis for heuristic search.
- **Closed - loop three - component system**:
- **Search tree**: Select the edges most likely to appear on the final path for exploration.
- **Fuzzy collision detection**: Based on the concept of the safety region of the Jacobian matrix, check whether these edges will collide.
- **Heuristic tree**: Update the lower - bound estimate of the path cost according to the information fed back by the fuzzy collision detection module, guiding the search tree to focus more effectively on important edges.

#### 3. **Key technical contributions**
- **Heuristic search**: Derive heuristic information from the pre - computed roadmap to estimate the lower - bound of the path cost in a dynamic environment, instead of the commonly used L2 distance.
- **Closed - loop accelerated search**: Form a closed - loop system to ensure that only relevant areas are searched and reduce the number of unnecessary edge evaluations.
- **Jacobian matrix collision detection**: Develop a collision detection method based on the Jacobian matrix to accelerate the edge evaluation process and determine the safety region.

### Experimental verification
The experimental part shows the performance advantages of HIRO in the simulated and real - world, especially the speed improvement when dealing with dynamic environmental changes. The experimental results show that HIRO can be significantly faster than other benchmark methods (such as RRT, PRM, etc.) while maintaining a collision - free path, and can also perform well without prior knowledge of environmental information.

### Summary
HIRO solves the problem of quickly planning a collision - free path for robots in a dynamic environment by combining the pre - computed static environment roadmap and rapid online adjustment in a dynamic environment. This method not only improves the efficiency of path planning, but also ensures the safety and feasibility of the path.