WayFAST: Navigation With Predictive Traversability in the Field

被引:35
作者
Gasparino, Mateus, V [1 ]
Sivakumar, Arun N. [1 ]
Liu, Yixiao [1 ]
Velasquez, Andres E. B. [1 ]
Higuti, Vitor A. H. [2 ]
Rogers, John [3 ]
Huy Tran [4 ]
Chowdhary, Girish [1 ]
机构
[1] Univ Illinois Urbana Champaign UIUC, Field Robot Engn & Sci Hub FRESH, Champaign, IL 61801 USA
[2] Univ Illinois Urbana Champaign UIUC, EarthSense Inc, Champaign, IL 61801 USA
[3] DEVCOM Army Res Lab, Adelphi, MD 20783 USA
[4] Univ Illinois Urbana Champaign UIUC, Dept Aerosp Engn, Champaign, IL 61801 USA
基金
美国国家科学基金会;
关键词
Field robots; semantic scene understanding; vision-based navigation; PATH-INTEGRAL CONTROL; ROBOT NAVIGATION;
D O I
10.1109/LRA.2022.3193464
中图分类号
TP24 [机器人技术];
学科分类号
080202 ; 1405 ;
摘要
We present a self-supervised approach for learning to predict traversable paths for wheeled mobile robots that require good traction to navigate. Our algorithm, termed WayFAST (Waypoint Free Autonomous Systems for Traversability), uses RGB and depth data, along with navigation experience, to autonomously generate traversable paths in outdoor unstructured environments. Our key inspiration is that traction can be estimated for rolling robots using kinodynamic models. Using traction estimates provided by an online receding horizon estimator, we are able to train a traversability prediction neural network in a self-supervised manner, without requiring heuristics utilized by previous methods. We demonstrate the effectiveness of WayFAST through extensive field testing in varying environments, ranging from sandy dry beaches to forest canopies and snow covered grass fields. Our results clearly demonstrate that WayFAST can learn to avoid geometric obstacles as well as untraversable terrain, such as snow, which would be difficult to avoid with sensors that provide only geometric data, such as LiDAR. Furthermore, we show that our training pipeline based on online traction estimates is more data-efficient than other heuristic-based methods.
引用
收藏
页码:10651 / 10658
页数:8
相关论文
共 39 条
[21]   Moving Horizon Estimation for Mobile Robots With Multirate Sampling [J].
Liu, Andong ;
Zhang, Wen-An ;
Chen, Michael Z. Q. ;
Yu, Li .
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, 2017, 64 (02) :1457-1467
[22]   DroNet: Learning to Fly by Driving [J].
Loquercio, Antonio ;
Maqueda, Ana I. ;
del-Blanco, Carlos R. ;
Scaramuzza, Davide .
IEEE ROBOTICS AND AUTOMATION LETTERS, 2018, 3 (02) :1088-1095
[23]  
McCormac John, 2017, 2017 IEEE International Conference on Robotics and Automation (ICRA), P4628, DOI 10.1109/ICRA.2017.7989538
[24]   Learning robust perceptive locomotion for quadrupedal robots in the wild [J].
Miki, Takahiro ;
Lee, Joonho ;
Hwangbo, Jemin ;
Wellhausen, Lorenz ;
Koltun, Vladlen ;
Hutter, Marco .
SCIENCE ROBOTICS, 2022, 7 (62)
[25]   The Artemis Jr. rover: Mobility platform for lunar ISRU mission simulation [J].
Reid, Ewan ;
Iles, Peter ;
Muise, Jason ;
Cristello, Nick ;
Jones, Brad ;
Faragalli, Michele ;
Visscher, Peter ;
Boucher, Dale ;
Simard-Bilodeau, Vincent ;
Apostolopoulos, Dimi ;
Rocco, Paul ;
Picard, Martin .
ADVANCES IN SPACE RESEARCH, 2015, 55 (10) :2472-2483
[26]  
Siegwart R., 2011, INTRO AUTONOMOUS MOB, DOI DOI 10.5860/CHOICE.49-1492
[27]  
Sivakumar AN, 2021, ROBOT SCI SYS
[28]   Real-Time Fusion Network for RGB-D Semantic Segmentation Incorporating Unexpected Obstacle Detection for Road-Driving Images [J].
Sun, Lei ;
Yang, Kailun ;
Hu, Xinxin ;
Hu, Weijian ;
Wang, Kaiwei .
IEEE ROBOTICS AND AUTOMATION LETTERS, 2020, 5 (04) :5558-5565
[29]   Tracking of Unicycle Robots Using Event-Based MPC With Adaptive Prediction Horizon [J].
Sun, Zhongqi ;
Xia, Yuanqing ;
Dai, Li ;
Campoy, Pascual .
IEEE-ASME TRANSACTIONS ON MECHATRONICS, 2020, 25 (02) :739-749
[30]   Probabilistic robotics [J].
Thrun, S .
COMMUNICATIONS OF THE ACM, 2002, 45 (03) :52-57