Abstract:Highlights • An autonomous exploration method for a car-like mobile robot on the ground is designed. Even if the degree of freedom of exploration on the ground is limited and the process is more difficult, this method expands the practical application of autonomous exploration. • The method is an interconnected and independent framework, including the exploration and the obstacle avoidance levels, both of which are dynamically expanded in the re-planning step, and finally realize the efficient exploration of the unknown environment. • The purpose of the exploration level is to provide the robot with a clear exploration direction, so that the robot can move to the unknown area as much as possible, improve the efficiency of the robot in exploring the unknown space, and effectively save computing resources. • The obstacle avoidance level processes the robot's local environment in real time. Based on the principle that the environment affects the robot's motion state, it provides the robot with exploration power and ensures the safety of the exploration process. Achieving efficient and safe autonomous exploration in unknown environments is an urgent challenge to be overcome in the field of robotics. Existing exploration methods based on random and greedy strategies cannot ensure that the robot moves to the unknown area as much as possible, and the exploration efficiency is not high. In addition, because the robot is located in an unknown environment, the robot cannot obtain enough information to process the surrounding environment and cannot guarantee absolute safety. To improve the efficiency and safety of exploring unknown environments, we propose an autonomous exploration motion planning framework that is divided into the exploration and obstacle avoidance levels. The two levels are independent and interconnected. The exploration level finds the optimal frontier target point in the global scope based on the forward filtering angle and cost function, attracting the robot to move to the unknown area as much as possible, and improving the exploration efficiency; the obstacle avoidance level establishes a scenario-speed conversion mechanism, and the target point and obstacle information are weighed to realise dynamic motion planning and completes obstacle avoidance control, and ensures the safety of exploration. Experiments in different simulation scenarios and real environments verify the superiority of the method. Results show that our method is superior to the existing methods.

A Hierarchical Approach for Mobile Robot Exploration in Pedestrian Crowd

CrowdMove: Autonomous Mapless Navigation in Crowded Scenarios

Efficient Exploration in Crowds by Coupling Navigation Controller and Exploration Planner

Real-Time Path Planning for Mobile Robots

Crowd-Aware Robot Navigation for Pedestrians with Multiple Collision Avoidance Strategies via Map-based Deep Reinforcement Learning

Robot Navigation in Crowd Based on Dual Social Attention Deep Reinforcement Learning

Robot navigation in a crowd by integrating deep reinforcement learning and online planning

Real-Time Navigation In Dynamic Human Environments Using Optimal Reciprocal Collision Avoidance

Deep Reinforcement Learning-Based Mapless Crowd Navigation with Perceived Risk of the Moving Crowd for Mobile Robots

A Hierarchical Autonomous Exploration Algorithm for Large-scale and Complex Environments with Mobile Robot.

Crowd-Aware Socially Compliant Robot Navigation via Deep Reinforcement Learning

SafeCrowdNav: safety evaluation of robot crowd navigation in complex scenes

Robot navigation in crowds via deep reinforcement learning with modeling of obstacle uni-action

Tactile Aware Dynamic Obstacle Avoidance in Crowded Environment with Deep Reinforcement Learning

Hierarchical framework for mobile robots to effectively and autonomously explore unknown environments

Safe and socially compliant robot navigation in crowds with fast-moving pedestrians via deep reinforcement learning

Robot Crowd Navigation Based on Spatio-Temporal Interaction Graphs and Danger Zones

Social crowd navigation of a mobile robot based on human trajectory prediction and hybrid sensing

A Novel Autonomous Exploration Algorithm Via LiDAR/IMU SLAM and Hierarchical Subsystem for Mobile Robot in Unknown Indoor Environments

Risk-Aware Deep Reinforcement Learning for Robot Crowd Navigation

Minimally-intrusive Navigation in Dense Crowds with Integrated Macro and Micro-level Dynamics