Efficient Trajectory Generation Based on Traversable Planes in 3D Complex Architectural Spaces

TL;DR

The paper presents an efficient trajectory generation strategy for ground robots operating in complex 3D architectural spaces by extracting traversable planes from point clouds to form a lightweight plane graph. It combines plane-graph path searching with cross-plane MS trajectory optimization, incorporating ALM-based constraints, final-position consistency, and safety/orientation requirements to produce smooth, safe motions across multiple layers. The approach is validated through simulations and a real-world CubeTrack deployment, demonstrating faster planning, reduced map size, and robust handling of stairs and ramps. This plane-centric framework enables scalable, high-quality trajectory generation for large-scale architectural environments with practical impact on autonomous navigation in multi-layer facilities.

Abstract

With the increasing integration of robots into human life, their role in architectural spaces where people spend most of their time has become more prominent. While motion capabilities and accurate localization for automated robots have rapidly developed, the challenge remains to generate efficient, smooth, comprehensive, and high-quality trajectories in these areas. In this paper, we propose a novel efficient planner for ground robots to autonomously navigate in large complex multi-layered architectural spaces. Considering that traversable regions typically include ground, slopes, and stairs, which are planar or nearly planar structures, we simplify the problem to navigation within and between complex intersecting planes. We first extract traversable planes from 3D point clouds through segmenting, merging, classifying, and connecting to build a plane-graph, which is lightweight but fully represents the traversable regions. We then build a trajectory optimization based on motion state trajectory and fully consider special constraints when crossing multi-layer planes to maximize the robot's maneuverability. We conduct experiments in simulated environments and test on a CubeTrack robot in real-world scenarios, validating the method's effectiveness and practicality.

Figures (8)

• Figure 1: The proposed trajectory generation enables the robot to navigate in a complex environment, passing through high platforms via stairs and slopes to avoid an obstacle wall and finally arrive at the target in the map.
• Figure 2: The process of extracting traversable planes. (a) Original point clouds. (b) Planes extracted using the region-growing. (c) Merged traversable planes. (d) Connectivity between traversable planes, where white lines are intersection lines. (e) Gridding. (f) ESDF.
• Figure 3: Grid states. Considering the expanded boundaries, the grid map is larger than the plane itself. The top left corner shows the model.
• Figure 4: Simulation environment and trajectory generation of the proposed method. The first row shows the point clouds of the environment: Planes(a1), Platform(b1), Multi-layer(c1), and Building(d1). The second row shows traversable planes, where white lines are intersection lines between planes, red lines are the plane graph, and green lines are the searched path. The third row shows ESDF and the generated trajectory. The table shows the trajectory length $L$ and trajectory optimization time $T_o$. The proposed method is capable of generating feasible and safe trajectories in complex architectural spaces.
• Figure 5: The trajectory of going upstairs and downstairs. The trajectory has a higher velocity when moving straight on the horizontal plane or going downstairs compared to the lower velocity when going upstairs. The trajectory switches to the next plane at the intersection line, ensuring velocity continuity during the switch.