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

Ze Wang, Zhenyu Gao, Jingang Qu, Pascal Morin · 2025

Read Histo-Planner: A Real-time Local Planner for MAVs Teleoperation based on Histogram of Obstacle Distribution by Ze Wang, Zhenyu Gao, Jingang Qu, Pascal Morin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-15043.txt.

Introduction

Ze Wang and co-authors present a real-time planner for small flying vehicles in a paper from two thousand twenty-five. The planner helps a human operator steer a micro air vehicle through cluttered spaces. It must avoid obstacles without building a full map of the surroundings, and with limited onboard computing.

The method builds a histogram of where obstacles lie around the vehicle and plans a path from that summary. A manager switches planning modes based on obstacle positions. The team validated the system in simulations and indoor flights for a teleoperation task. They also compared it against other planners on a purpose-built simulation platform.

Catalog record

CREATORZe Wang, Zhenyu Gao, Jingang Qu, Pascal Morin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE30 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-15043

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Histo-Planner: A Real-time Local Planner for MAVs Teleoperation based on Histogram of Obstacle Distribution Ze Wang, Zhenyu Gao, Jingang Qu, Pascal Morin 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. I INTRODUCTION Micro aerial vehicles (MAVs) are used in many applications, such as rescue search, forestry monitoring, infrastructure maintenance, aerial photography, etc. When the MAV operates in cluttered environments, obstacle avoidance is a major problem. Solutions to this problem are highly dependent on the type of environment, the available onboard sensors, the availability of a global map of the environment, and the available computational power. While solutions to this problem rely on both perception and planning/navigation aspects (the classical sense and avoid scenario), the present paper focuses on the navigation aspect. Many traditional navigation methods are summarized in detail in [ 1 ] . They can be divided into three categories, namely i) Graph-based discrete path planning algorithms: A ∗ , RRT ∗ (Rapidly-exploring random tree), PRM (Probabilistic roadmap), Visibility Roadmap, ii) Gradient-Field-based continuous path planning algorithms: Artificial Potential Field (APF) [ 2 ] , Harmonic potential field [ 3 ] and Gradient-based Trajectory Optimization (GTO), etc., and iii) Goal-conscious Local obstacle avoidance algorithms: Vector Field Histogram (VFH) [ 4 ] [ 5 ] , Dynamic window approaches (DW) [ 6 ] , Geometric Guidance (GG) [ 7 ] , Streamline methods [ 8 ] etc. Local obstacle avoidance algorithms [ 1 ] are also called ”path searching algorithms”. Fig. 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. Usually navigation maps are divided into free space, occupied space and unknown space. Due to its completeness, the discrete path planning algorithm represented by A* is well suited for scenarios where the global map is known. When a global map is not available, people usually prefer frontier-based algorithms [ 9 ] for space exploration tasks, which are also regarded as a local obstacle avoidance algorithm that focuses on the local obstacles. Continuous path planning algorithms can generate a trajectory with high-order continuity to be tracked by the MAV, but this algorithm often falls into local minima. Harmonic potential field, inspired by fluid mechanics, skillfully avoids local minima but introduces a large amount of calculation. However, continuous path planning methods, such as gradient-based algorithms, can be combined with frontier based methods where the latter help avoiding local minima. Thus, combining a continuous trajectory planning algorithm with a goal-conscious local obstacle avoidance algorithm can be used as a robust local trajectory planner avoiding local minima and obstacles. Once a first collision-free trajectory has been obtained, a replanning is often performed. Recent related works [ 10 ] - [ 15 ] reveal that gradient-based trajectory optimization approaches typically formulates trajectory replanning as an optimization problem. Different methods of trajectory description have been used in these works such as polynomial curves [ 10 ] , [ 11 ] , Bézier curves [ 12 ] , or B-spline curves [ 13 ] , [ 14 ] , [ 15 ] , [ 16 ] . [ 14 ] provides that the change of an individual control point of B-splines only affects the nearest curve segments, which is important to generate a reliable trajectory. An approach that avoids the need of ESDF and only relies on the occupancy grid is proposed in [ 15 ] . Differently, we aim to generate local obstacle avoidance trajectories for non-globally known environments while avoiding global static occupancy grid updates and saving further computation time. In this paper, we propose a ”guided” gradient-based optimization trajectory. Each time trajectory planning is necessary, a ”guidance point” is generated based on the local obstacle distribution, the MAV’s state, and the goal location. This guidance point allows us to define an initial trajectory for the optimizer, that is helpful to avoid the local minima problem. It is generated based on a histogram of local obstacles that avoids the need to update an occupancy grid, thus saving significant computation time. Then, an optimizer is applied to generate the MAV’s reference trajectory. The action of the optimizer can be seen as someone releasing an elastic rope with fixed end points from the guidance point. Additionally, a planner manager is proposed to trigger different navigation modes based on the goal point and the MAV’s current location with respect to the obstacles. The paper is organized as follows. Sect. II gives an overview of the proposed navigation architecture. Sect. III discusses the Histogram of Obstacle Distribution generation and guidance point generation. Sect. IV describes the gradient-based trajectory optimization and Sect. V describes the planner manager. Simulation and experimental results are reported in Sect. VI , together with benchmark comparisons. II Overview Fig. 2 shows the architecture of the proposed navigation system. The system is divided into three main parts: the histogram initialization and guidance point generation detailed in Sec. III , the optimizer detailed in Sec. IV , and the planner manager detailed in Sec. V . The navigation system performs in the order marked in the diagram. Input Information: The system uses as inputs the current point clouds of the local environment, estimated MAV’s states and goal point. Point clouds are typically obtained from depth camera or lidars. MAV’s states contain the flight speed and position of the vehicle in local frame. The goal point can be specified by different means, depending on the information provided to the pilot. For teleoperation applications, goal point may be provided through a ground station, or even directly with a joystick. Histogram and Guidance point generation: The navigation system starts with the histogram generation. The obstacle histogram is updated as a separate thread in real time from the received sensor point clouds. The histogram is used to describe the local obstacle space by a limited number of point clouds, providing obstacle distance monitoring and local gradient field generation when needed. The histogram has an image-like data structure and thus nearby obstacles have a higher resolution. Once the trajectory generation loop is activated, the obstacle histogram inflates the obstacle space according to the safety distance, and combines information of current flight states and goal location in order to generate a weighted histogram. Leveraging image processing methods, a filter kernel acting on the weighted histogram is used to find the best obstacle gap leading to the goal, thereby generating the ”guidance point”. Fig. 2 : A global view of the navigation system architecture. Optimized trajectory: Reference trajectories for the MAV are defined as B-splines, with associated control points. The location of the

Read the complete text