ColorDynamic: Generalizable, Scalable, Real-time, End-to-end Local Planner for Unstructured and Dynamic Environments

TL;DR

ColorDynamic introduces an end-to-end local planning framework that directly maps temporal lidar observations to robot commands in unstructured and dynamic environments. It advances temporal perception with Transqer, a Transformer-based processor using a TWQ, and enhances training with E-Sparrow’s PGVD and Symmetric Invariance augmentation, achieving real-time planning and strong generalization. Integrated into the OPCD system with OkayPlan, ColorDynamic demonstrates high safety and efficiency in both simulated and real-world deployments, with reported success rates above $0.90$ and planning times around $1.2$–$1.3$ ms, and it is open-sourced for reproducibility. The work shows significant potential for scalable, multi-robot deployment in complex environments, while suggesting extensions to 3D navigation for broader applicability.

Abstract

Deep Reinforcement Learning (DRL) has demonstrated potential in addressing robotic local planning problems, yet its efficacy remains constrained in highly unstructured and dynamic environments. To address these challenges, this study proposes the ColorDynamic framework. First, an end-to-end DRL formulation is established, which maps raw sensor data directly to control commands, thereby ensuring compatibility with unstructured environments. Under this formulation, a novel network, Transqer, is introduced. The Transqer enables online DRL learning from temporal transitions, substantially enhancing decision-making in dynamic scenarios. To facilitate scalable training of Transqer with diverse data, an efficient simulation platform E-Sparrow, along with a data augmentation technique leveraging symmetric invariance, are developed. Comparative evaluations against state-of-the-art methods, alongside assessments of generalizability, scalability, and real-time performance, were conducted to validate the effectiveness of ColorDynamic. Results indicate that our approach achieves a success rate exceeding 90% while exhibiting real-time capacity (1.2-1.3 ms per planning). Additionally, ablation studies were performed to corroborate the contributions of individual components. Building on this, the OkayPlan-ColorDynamic (OPCD) navigation system is presented, with simulated and real-world experiments demonstrating its superiority and applicability in complex scenarios. The codebase and experimental demonstrations have been open-sourced on our website to facilitate reproducibility and further research.

Figures (17)

• Figure 1: Illustration of $D2T$, $\alpha$, and $r_o$.
• Figure 2: A comparison between Sparrow (a) and E-Sparrow (b)$\sim$(f). The purple circles represent the robots, while the blue fan-shaped lines indicate the lidar beams. The black lumps and red circles denote static and dynamic obstacles, respectively. The green-framed regions highlight the target points. In the multi-robot scenarios (e) and (f), to distinguish the target points of different robots, their colors are aligned with robots' external contours. It is noteworthy that in E-Sparrow, obstacles are randomly generated by the program, and dynamic obstacles move randomly within the environment.
• Figure 3: Architecture of the Transqer.
• Figure 4: Symmetric Invariance.
• Figure 5: Diagram of OkayPlan.