|
Autonomous Navigation System with Obstacle Avoidance using 2.5D Map Generated by Point Cloud |
|
Haeyeon Gim(POSTECH, Korea), Minwook Jeong, Sohee Han(Postech, Korea) |
| In this paper, we propose an autonomous navigation system of a mobile robot in a dynamic environment. We build a 2.5D map, that integrates a 2D grid map with dynamic objects 3D geometry information. From the 3D LiDAR point cloud, dynamic points are detected by tracking the occupancy changes over time. Then remained static points are used for generating a 2D grid map by SLAM algorithm. A computation cost is reduced efficiently by reconstructing only the necessary parts for the mobile robot driving into the high resolution of raw point cloud data. |
|
|
Integration of Path Optimization and Obstacle avoidance for Autonomous Precision Immobilization Technique Maneuver |
|
Rahul Meel, Rohit Kumar(Indian Institute Of Technology, Guwahati, India) |
| The PIT (Precision Immobilization Technique) maneuver is a pursuit tactic used by law-enforcement officials to handle dangerous vehicle pursuit situations, wherein a pursuing car can force a fleeing car to lose control and stop. Autonomous PIT maneuvers could be the next step in handling dangerous vehicle pursuits of fugitives. This work presents an extension in the broader Autonmous PIT maneuver field by implementation of path optimization (to ensure that the bullet vehicle arrives to the target vehicle following the optimized path) along with obstacle avoidance in physical environment for execution of autonomous PIT maneuvering. |
|
|
Fast, Lightweight, and Robust Path Planning for Low-power Embedded Environments |
|
Hwanhee Lee, Ilyong Yoon, Sangwoo Kim, Dong Jin Hyun(Hyundai Motor Company, Korea) |
| A fast, lightweight, and robust path planning algorithm is proposed for single agent robot navigation environments varying every moments, and its feasibility for low-power embedded environments are verified.
Main objective of this paper is to find path immediately avoiding obstacles in unstructured environments for safe robot navigation on low-power embedded environments. Several experiments are conducted to evaluate the path planning performance and the long-term operation and to verify the consistency of performance and the feasibility for low-power embedded environments. |
|
|
Heuristic Guided Artificial Potential Field for Avoidance of Small Obstacles |
|
Sagar Dalai, Mahammad Irfan, Samarth Singh, Kaushal Kishore, S.A Akbar(Council of Scientific & Industrial Research, India) |
| The work presents a heuristic guided Artificial Potential Field(APF) based algorithm to find a practical trajectory for an Autonomous Unmanned Aerial Vehicle (UAV) path planning avoiding the issue of local minima. The proposed algorithm is tested and validated against existing general potential field techniques for different simulation scenarios in a 3D simulated environment using ROS and Gazebo supported PX4-SITL. |
|
|
Encoding A Mathematically Faithful DeepVIO Solution |
|
Amar Ali N. Khan(City, University of London, United Kingdom), Nabil Aouf(City University of London, United Kingdom) |
| In order to enable Deep Learning (DL) based methods to be better understood, without, the requirement of extensive probing, this paper present a method by which models are forced to comply with a 100% faithfulness to existing mathematical solutions. Whilst the pre-existing solutions can be completely understood, they do not always provide the desired results. This is often due to an inability to tune hyperparameters such as feature detection kernels. By mapping the existing solutions to DL models, optimisation of such parameters can be achieved via backpropagation. This paper presents one such method for the task of visual odometry (6 DoF pose estimation). |
| |