Stanley: The Robot That Won the DARPA Grand Challenge

Thrun, Sebastian; Montemerlo, Michael; Dahlkamp, Hendrik; Stavens, David; Aron, Andrei; Diebel, James; Fong, Philip W. L.; Gale, John T.; Halpenny, Morgan; Hoffmann, Gabriel; Lau, Kenny; Oakley, Celia M.; Palatucci, Mark; Pratt, Vaughan R.; Stang, Pascal; Strohband, Sven; Dupont, Cedric; Jendrossek, Lars-Erik; Koelen, Christian; Markey, Charles; Rummel, Carlo; Niekerk, Joe van; Jensen, Eric L. N.; Alessandrini, Philippe; Bradski, Gary; Davies, Bob; Ettinger, Scott; Kaehler, Adrian; Nefian, Ara; Mahoney, Pamela

doi:10.1007/978-3-540-73429-1_1

Cited by 373 publications

(422 citation statements)

References 17 publications

Supporting

Mentioning

415

Contrasting

Unclassified

Order By: Relevance

“…Since a short perceptual horizon limits the available stopping distance of the robot, the map itself is insufficient for high-speed navigation. As has been shown previously, using learned models for image-based guidance can significantly extend the effective perceptual range and improve navigation performance [13,34].…”

Section: Visual Navigation Problem Formulationmentioning

confidence: 68%

Safe Visual Navigation via Deep Learning and Novelty Detection

Richter

Roy

2017

Robotics: Science and Systems XIII

174

View full text Add to dashboard Cite

Abstract-Robots that use learned perceptual models in the real world must be able to safely handle cases where they are forced to make decisions in scenarios that are unlike any of their training examples. However, state-of-the-art deep learning methods are known to produce erratic or unsafe predictions when faced with novel inputs. Furthermore, recent ensemble, bootstrap and dropout methods for quantifying neural network uncertainty may not efficiently provide accurate uncertainty estimates when queried with inputs that are very different from their training data. Rather than unconditionally trusting the predictions of a neural network for unpredictable real-world data, we use an autoencoder to recognize when a query is novel, and revert to a safe prior behavior. With this capability, we can deploy an autonomous deep learning system in arbitrary environments, without concern for whether it has received the appropriate training. We demonstrate our method with a vision-guided robot that can leverage its deep neural network to navigate 50% faster than a safe baseline policy in familiar types of environments, while reverting to the prior behavior in novel environments so that it can safely collect additional training data and continually improve. A video illustrating our approach is available at: http://groups.csail.mit.edu/rrg/videos/safe visual navigation.

show abstract

Section: Visual Navigation Problem Formulationmentioning

confidence: 68%

Safe Visual Navigation via Deep Learning and Novelty Detection

Richter

Roy

2017

Robotics: Science and Systems XIII

174

View full text Add to dashboard Cite

show abstract

“…Moreover, LIDAR mapping systems are able to rapidly acquire large-scale 3-D point cloud data for real-time vision, with jointly providing accurate 3-D geometrical information of the scene, and additional features about the reflection properties and compactness of the surfaces. The detection of urban objects is a fundamental problem in any perception motivated point cloud processing task [15]. Although it is a challenging problem itself, it can be helpful for several robot vision tasks, such as object recognition, localization or feature extraction.…”

Section: Problem Statementmentioning

confidence: 99%

Fast 3-D Urban Object Detection on Streaming Point Clouds

Börcs

Nagy

Benedek

2015

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Efficient and fast object detection from continuously streamed 3-D point clouds has a major impact in many related research tasks, such as autonomous driving, self localization and mapping and understanding large scale environment. This paper presents a LIDAR-based framework, which provides fast detection of 3-D urban objects from point cloud sequences of a Velodyne HDL-64E terrestrial LIDAR scanner installed on a moving platform. The pipeline of our framework receives raw streams of 3-D data, and produces distinct groups of points which belong to different urban objects. In the proposed framework we present a simple, yet efficient hierarchical grid data structure and corresponding algorithms that significantly improve the processing speed of the object detection task. Furthermore, we show that this approach confidently handles streaming data, and provides a speedup of two orders of magnitude, with increased detection accuracy compared to a baseline connected component analysis algorithm.

show abstract

“…[9]) where there are physical, man-made structures that assist in the registration of 3d scans, improving the quality of the resulting maps. Other examples include mapping solutions for off-road autonomous car driving [13], mining operations [5] and autonomous road inspection [8]. The latter system (RoadBot) fuses information from range scanners thanks to precise pose estimation obtained from high precision GPS (i.e.…”

Section: Related Workmentioning

confidence: 99%

Evaluation of Laser Range-Finder Mapping for Agricultural Spraying Vehicles

Moreno

Cielniak

Duckett

2014

Towards Autonomous Robotic Systems

View full text Add to dashboard Cite

Abstract. In this paper, we present a new application of laser rangefinder sensing to agricultural spraying vehicles. The current generation of spraying vehicles use automatic controllers to maintain the height of the sprayer booms above the crop. However, these control systems are typically based on ultrasonic sensors mounted on the booms, which limits the accuracy of the measurements and the response of the controller to changes in the terrain, resulting in a sub-optimal spraying process. To overcome these limitations, we propose to use a laser scanner, attached to the front of the sprayer's cabin, to scan the ground surface in front of the vehicle and to build a scrolling 3d map of the terrain. We evaluate the proposed solution in a series of field tests, demonstrating that the approach provides a more detailed and accurate representation of the environment than the current sonar-based solution, and which can lead to the development of more efficient boom control systems.

show abstract

Stanley: The Robot That Won the DARPA Grand Challenge

Cited by 373 publications

References 17 publications

Safe Visual Navigation via Deep Learning and Novelty Detection

Safe Visual Navigation via Deep Learning and Novelty Detection

Fast 3-D Urban Object Detection on Streaming Point Clouds

Evaluation of Laser Range-Finder Mapping for Agricultural Spraying Vehicles

Contact Info

Product

Resources

About