EGO-Planner: An ESDF-free Gradient-based Local Planner for Quadrotors (2008.08835v2)

Abstract: Gradient-based planners are widely used for quadrotor local planning, in which a Euclidean Signed Distance Field (ESDF) is crucial for evaluating gradient magnitude and direction. Nevertheless, computing such a field has much redundancy since the trajectory optimization procedure only covers a very limited subspace of the ESDF updating range. In this paper, an ESDF-free gradient-based planning framework is proposed, which significantly reduces computation time. The main improvement is that the collision term in the penalty function is formulated by comparing the colliding trajectory with a collision-free guiding path. The resulting obstacle information will be stored only if the trajectory hits new obstacles, making the planner only extract necessary obstacle information. Then, we lengthen the time allocation if dynamical feasibility is violated. An anisotropic curve fitting algorithm is introduced to adjust higher-order derivatives of the trajectory while maintaining the original shape. Benchmark comparisons and real-world experiments verify its robustness and high-performance. The source code is released as ROS packages.

• The paper introduces an ESDF-free framework that reduces up to 70% of computation time by directly evaluating gradient information from collisions.
• The paper employs an anisotropic curve fitting algorithm to optimize trajectory smoothness and adjust time allocation when dynamic constraints are violated.
• The paper validates its approach through real-world experiments, demonstrating efficient and robust quadrotor navigation at high velocities in complex environments.

Overview of EGO-Planner: An ESDF-free Gradient-based Local Planner for Quadrotors

The paper introduces the EGO-Planner, a gradient-based local planner for quadrotors that operates without relying on an Euclidean Signed Distance Field (ESDF). The primary motivation for this development is the realization that the computation of ESDF, traditionally used in many existing gradient-based planning frameworks, constitutes up to 70% of the processing time during trajectory planning. This necessitates an approach that minimizes computational redundancy, specifically for applications constrained by resource capacities.

Key Contributions

1. ESDF-free Framework: EGO-Planner replaces the traditional ESDF structure by directly evaluating gradient information from the trajectory's interaction with obstacles. It employs a penalty function that assesses collisions by contrasting the trajectory with a pre-determined collision-free reference path. The obstacle information is dynamically updated and retained only when encountering new obstacles, significantly improving computational efficiency.
2. Trajectory Optimization and Feasibility: The proposed planner implements an anisotropic curve fitting algorithm that preserves the geometric properties of the trajectory while optimizing higher-order derivatives to ensure smoothness. In cases where dynamic feasibility constraints are violated, the planner adjusts time allocation without modifying trajectory shape conformity.
3. Real-world Evaluation: Benchmarks and experimental results validate the planner's capability in producing robust, high-performance quadrotor trajectories in both simulation and reality. Compared to ESDF-dependent methods, EGO-Planner achieves a similar success rate at a fraction of the computation time, highlighting its practical advantages for real-time applications.

Experimental Insights

EGO-Planner's performance was juxtaposed with ESDF-based frameworks, notably Fast-Planner and EWOK, across various obstacle densities. EGO-Planner generated smoother and shorter trajectories but at a marginally higher energy cost. The absence of an upfront requirement for a collision-free initialization, a significant challenge in ESDF approaches, emphasizes the planner's robustness in uncertain environments. Additionally, the evaluation indicates significant savings in computation time while ensuring collision-free navigation at high velocities, a vital parameter for real-time applications.

Implications and Future Work

The research presents a pioneering effort in ESDF-independent maneuvering for quadrotors, promising substantial implications in aerial robotics. Practically, this could facilitate wider autonomous deployment in computationally-demanding environments, extending applications across logistics, surveillance, and remote sensing. Theoretically, EGO-Planner challenges traditional reliance on ESDF for trajectory planning, encouraging exploration into alternative obstacle disparity metrics and optimization paradigms.

Future advancements could explore:

• Enhanced topological path exploration to overcome local minima in complex spaces.
• Dynamic environment adaptability by integrating real-time moving object recognition and path-replanning capabilities.
• Further optimization of the gradient evaluation process to reduce energy consumption while maintaining performance robustness.

In conclusion, EGO-Planner represents a noteworthy stride towards optimizing autonomous flight by delivering a computationally light and agile trajectory planning solution that does not compromise on safety or performance. Such innovations will undoubtedly influence future research directions in aerial robotics and autonomous navigation systems.