Histo-Planner: A Real-time Local Planner for MAVs Teleoperation based on Histogram of Obstacle Distribution

TL;DR

The paper tackles real-time obstacle avoidance for MAV teleoperation without relying on a global map. It introduces a histogram-based local planner that generates a guidance point from obstacle distribution and optimizes a $3$rd-degree B-spline trajectory via a gradient-based NLopt solver, with a planner manager for safety and mode switching. Key contributions include the histogram-based obstacle representation, weighted guidance point generation, and a robust planning architecture that supports Normal, Straightforward, and escape modes, validated through simulations and indoor experiments against Fast-Planner and Ego-Planner baselines. The approach enables efficient, map-free navigation in unknown cluttered environments, with practical potential for teleoperation in limited-compute platforms.

Abstract

This paper concerns real-time obstacle avoidance for micro aerial vehicles (MAVs). Motivated by teleoperation applications in cluttered environments with limited computational power, we propose a local planner that does not require the knowledge or construction of a global map of the obstacles. The proposed solution consists of a real-time trajectory planning algorithm that relies on the histogram of obstacle distribution and a planner manager that triggers different planning modes depending on obstacles location around the MAV. The proposed solution is validated, for a teleoperation application, with both simulations and indoor experiments. Benchmark comparisons based on a designed simulation platform are also provided.

Figures (7)

• Figure 1: The custom-built MAV flight platform of with a size of 15x20x20cm and weights about 500g. The MAV is equipped with Pixhawk 4 autopilot and Khadas Vim 3. All modules run on the on-board computer.
• Figure 2: A global view of the navigation system architecture.
• Figure 3: The dotted curve represents the initial trajectory, and the solid curve represents the optimized trajectory. The guidance point is generated from the histogram of obstacle distribution. The blue area is the obstacle area and the gray depicts its boundary.
• Figure 4: An obstacle appears in the field of view of the sensor. Measurements are projected onto a spherical coordinate system which is flattened into a Cartesian-like coordinate system.
• Figure 5: The right sub-figure features the vehicle's frame, guidance point (purple), initial trajectory (red), and optimal trajectory (yellow) based on the perceived environment. The obstacle histogram (upper-left sub-figure) is updated in real time based on the sensor information (color points), with the vertical axis representing the height direction and the horizontal axis representing the horizontal circumferential direction. The weighted histogram (lower-left sub-figure) is generated only if guidance point generation is required. The brightest spot represents the direction of the guidance point.