From Perception to Decision: A Data-driven Approach to End-to-end Motion Planning for Autonomous Ground Robots
arXiv:1609.07910 · doi:10.1109/ICRA.2017.7989182
Abstract
Learning from demonstration for motion planning is an ongoing research topic. In this paper we present a model that is able to learn the complex mapping from raw 2D-laser range findings and a target position to the required steering commands for the robot. To our best knowledge, this work presents the first approach that learns a target-oriented end-to-end navigation model for a robotic platform. The supervised model training is based on expert demonstrations generated in simulation with an existing motion planner. We demonstrate that the learned navigation model is directly transferable to previously unseen virtual and, more interestingly, real-world environments. It can safely navigate the robot through obstacle-cluttered environments to reach the provided targets. We present an extensive qualitative and quantitative evaluation of the neural network-based motion planner, and compare it to a grid-based global approach, both in simulation and in real-world experiments.
arXiv commentpresented at the IEEE International Conference on Robotics and Automation (ICRA) 2017
References in corpus (4)
Cited by in corpus (62)
- DRL-VO: Learning to Navigate Through Crowded Dynamic Scenes Using Velocity Obstacles
- Perception and Navigation in Autonomous Systems in the Era of Learning: A Survey
- Virtual-to-real Deep Reinforcement Learning: Continuous Control of Mobile Robots for Mapless Navigation
- Spatial Action Maps for Mobile Manipulation
- Intention-Net: Integrating Planning and Deep Learning for Goal-Directed Autonomous Navigation
- A Survey of Deep Network Solutions for Learning Control in Robotics: From Reinforcement to Imitation
- Deep Local Trajectory Replanning and Control for Robot Navigation
- Probabilistic End-to-End Vehicle Navigation in Complex Dynamic Environments with Multimodal Sensor Fusion
- Efficient Processing of Deep Neural Networks: A Tutorial and Survey
- CrowdMove: Autonomous Mapless Navigation in Crowded Scenarios
- iPlanner: Imperative Path Planning
- Towards Optimally Decentralized Multi-Robot Collision Avoidance via Deep Reinforcement Learning
- APPLE: Adaptive Planner Parameter Learning from Evaluative Feedback
- Cross-Modal Contrastive Learning of Representations for Navigation using Lightweight, Low-Cost Millimeter Wave Radar for Adverse Environmental Conditions
- Realtime Collision Avoidance for Mobile Robots in Dense Crowds using Implicit Multi-sensor Fusion and Deep Reinforcement Learning
- Learning a Decentralized Multi-arm Motion Planner
- The Emotionally Intelligent Robot: Improving Social Navigation in Crowded Environments
- CAHSOR: Competence-Aware High-Speed Off-Road Ground Navigation in SE(3)
- To Go or Not To Go? A Near Unsupervised Learning Approach For Robot Navigation
- Deep Network Uncertainty Maps for Indoor Navigation
- Curriculum Learning for Safe Mapless Navigation
- Human-Inspired Multi-Agent Navigation using Knowledge Distillation
- MAPPER: Multi-Agent Path Planning with Evolutionary Reinforcement Learning in Mixed Dynamic Environments
- Learning Obstacle Representations for Neural Motion Planning
- A Survey of Behavior Learning Applications in Robotics -- State of the Art and Perspectives
- Robot Navigation with Map-Based Deep Reinforcement Learning
- A Lifelong Learning Approach to Mobile Robot Navigation
- GONet: A Semi-Supervised Deep Learning Approach For Traversability Estimation
- Carl-Lead: Lidar-based End-to-End Autonomous Driving with Contrastive Deep Reinforcement Learning
- Neural Network Memory Architectures for Autonomous Robot Navigation
- Learning to Navigate: Exploiting Deep Networks to Inform Sample-Based Planning During Vision-Based Navigation
- APPLD: Adaptive Planner Parameter Learning from Demonstration
- Hallucinating robots: Inferring Obstacle Distances from Partial Laser Measurements
- Transformer-Enhanced Motion Planner: Attention-Guided Sampling for State-Specific Decision Making
- LCP: A Low-Communication Parallelization Method for Fast Neural Network Inference in Image Recognition
- Learning Navigation Behaviors End-to-End with AutoRL
- Learning Inverse Kinodynamics for Accurate High-Speed Off-Road Navigation on Unstructured Terrain
- Role Playing Learning for Socially Concomitant Mobile Robot Navigation
- Multi-modal Experts Network for Autonomous Driving
- Autonomous Navigation in Dynamic Environments: Deep Learning-Based Approach
- Agile Robot Navigation through Hallucinated Learning and Sober Deployment
- Real-time Multi-target Path Prediction and Planning for Autonomous Driving aided by FCN
- Avoidance of Manual Labeling in Robotic Autonomous Navigation Through Multi-Sensory Semi-Supervised Learning
- Rationally Inattentive Path-Planning via RRT*
- Programming and Deployment of Autonomous Swarms using Multi-Agent Reinforcement Learning
- Multi-criteria Evolution of Neural Network Topologies: Balancing Experience and Performance in Autonomous Systems
- Learning Resilient Behaviors for Navigation Under Uncertainty
- Distilling Motion Planner Augmented Policies into Visual Control Policies for Robot Manipulation
- APPLR: Adaptive Planner Parameter Learning from Reinforcement
- Hierarchical Reinforcement Learning Framework towards Multi-agent Navigation
- VUNet: Dynamic Scene View Synthesis for Traversability Estimation using an RGB Camera
- Fast-reactive probabilistic motion planning for high-dimensional robots
- The Pitfall of More Powerful Autoencoders in Lidar-Based Navigation
- OF-VO: Efficient Navigation among Pedestrians Using Commodity Sensors
- S2P2: Self-Supervised Goal-Directed Path Planning Using RGB-D Data for Robotic Wheelchairs
- Toward Agile Maneuvers in Highly Constrained Spaces: Learning from Hallucination
- Creating Robust Deep Neural Networks With Coded Distributed Computing for IoT Systems
- iWalker: Imperative Visual Planning for Walking Humanoid Robot
- Neural Flocking: MPC-based Supervised Learning of Flocking Controllers
- Machine Vision for Improved Human-Robot Cooperation in Adverse Underwater Conditions
- DeepMNavigate: Deep Reinforced Multi-Robot Navigation Unifying Local & Global Collision Avoidance
- Learning with Training Wheels: Speeding up Training with a Simple Controller for Deep Reinforcement Learning