Toward Reliable Off Road Autonomous Vehicles Operating in Challenging Environments

Kelly, Alonzo; Stentz, Anthony; Amidi, Omead; Bode, M. F.; Bradley, David M.; Diaz-Calderon, Antonio; Happold, Mike; Herman, Herman; Mandelbaum, Robert; Pilarski, Tom; Rander, Peter; Thayer, Scott; Vallidis, Nick; Warner, Randy

doi:10.1177/0278364906065543

Cited by 209 publications

(93 citation statements)

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Intuitively, we expect space containing vegetation to be seen as relatively "porous" by the LiDAR sensor, compared to space containing solid objects. Researchers have used this intuition to model vegetation by counting the ratio of LiDAR hits to pass-throughs [23] for each grid cell in a map. We use a probabilistic formulation of this idea, in which we interpret the posterior mean of a Bernoulli with a uniform Beta prior as the density of each cell.…”

Section: B Volumetric Density Mappingmentioning

confidence: 99%

3D Convolutional Neural Networks for landing zone detection from LiDAR

Maturana

Scherer

2015

2015 IEEE International Conference on Robotics and Automation (ICRA)

194

105

View full text Add to dashboard Cite

We present a system for the detection of small and potentially obscured obstacles in vegetated terrain. The key novelty of this system is the coupling of a volumetric occupancy map with a 3D Convolutional Neural Network (CNN), which to the best of our knowledge has not been previously done. This architecture allows us to train an extremely efficient and highly accurate system for detection tasks from raw occupancy data. We apply this method to the problem of detecting safe landing zones for autonomous helicopters from LiDAR point clouds. Current methods for this problem rely on heuristic rules and use simple geometric features. These heuristics break down in the presence of low vegetation, as they do not distinguish between vegetation that may be landed on and solid objects that should be avoided. We evaluate the system with a combination of real and synthetic range data. We show our system outperforms various benchmarks, including a system integrating various hand-crafted point cloud features from the literature.

show abstract

Section: B Volumetric Density Mappingmentioning

confidence: 99%

3D Convolutional Neural Networks for landing zone detection from LiDAR

Maturana

Scherer

2015

2015 IEEE International Conference on Robotics and Automation (ICRA)

194

105

View full text Add to dashboard Cite

show abstract

“…Kelly et al 8 describe the design and operation of a human-robot team for off-road navigation, wherein terrain classification is based on geometry-based features combined with multi-spectral image-based features. The robot-support surface is extracted by ray-tracing of laser-beams and training a neural network to derive the load-bearing surface when traversing over vegetated areas, while negative obstacles Fig.…”

Section: Previous Workmentioning

confidence: 99%

Mind the gap: detection and traversability analysis of terrain gaps using LIDAR for safe robot navigation

Sinha

Papadakis

2013

Robotica

View full text Add to dashboard Cite

SUMMARYSafe navigation of robotic vehicles is considered as a key pre-requisite of successful mission operations within highly adverse and unconstrained environments. While there has been extensive research in the perception of positive obstacles, little progress can be accredited to the field of negative obstacles. This paper hypostatizes an elaborative attempt to address the problem of negative obstacle detection and traversability analysis in the form of gaps by processing 3-dimensional range data. The domain of application concerns Urban Search and Rescue scenarios that reflect environments of increased complexity in terms of diverse terrain irregularities. To allow real-time performance and, in turn, timely prevention of unrecoverable robotic states, the proposed approach is based on the application of efficient image morphological operations for noise reduction and border following the detection and grouping of gaps. Furthermore, we reason about gap traversability, a concept that is novel within the field. Traversability assessments are based on features extracted through Principal Component Analysis by exploring the spatial distribution of the interior of the individual gaps or the orientation distribution of the corresponding contour. The proposed approach is evaluated within a realistic scenario of a tunnel car accident site and a challenging outdoor scenario. Using a contemporary Search and Rescue robot, we have performed extensive experiments under various parameter settings that allowed the robot to always detect the real gaps, and either optimally cross over those that were traversable or otherwise avoid them.

show abstract

“…(1) as the use of local and global reference frames in Bacha et al (2008), Kelly et al (2006), Leonard et al (2008), Miller et al (2008), and Moore et al (2009). The timing synchronization depends on a combination of hardware and software, such as using the Network Time Protocol (NTP) and pulse per second (PPS) signals (Mills, 2006;Network Time Protocol, 2008).…”

Section: Sources Of Errormentioning

confidence: 99%

Error modeling and calibration of exteroceptive sensors for accurate mapping applications

Underwood

Hill

Peynot

et al. 2009

Journal of Field Robotics

View full text Add to dashboard Cite

Reliable robotic perception and planning are critical to performing autonomous actions in uncertain, unstructured environments. In field robotic systems, automation is achieved by interpreting exteroceptive sensor information to infer something about the world. This is then mapped to provide a consistent spatial context, so that actions can be planned around the predicted future interaction of the robot and the world. The whole system is as reliable as the weakest link in this chain. In this paper, the term mapping is used broadly to describe the transformation of range-based exteroceptive sensor data (such as LIDAR or stereo vision) to a fixed navigation frame, so that it can be used to form an internal representation of the environment. The coordinate transformation from the sensor frame to the navigation frame is analyzed to produce a spatial error model that captures the dominant geometric and temporal sources of mapping error. This allows the mapping accuracy to be calculated at run time. A generic extrinsic calibration method for exteroceptive range-based sensors is then presented to determine the sensor location and orientation. This allows systematic errors in individual sensors to be minimized, and when multiple sensors are used, it minimizes the systematic contradiction between them to enable reliable multisensor data fusion. The mathematical derivations at the core of this model are not particularly novel or complicated, but the rigorous analysis and application to field robotics seems to be largely absent from the literature to date. The techniques in this paper are simple to implement, and they offer a significant improvement to the accuracy, precision, and integrity of mapped information. Consequently, they should be employed whenever maps are formed from range-based exteroceptive sensor data. C 2009 Wiley Periodicals, Inc.

show abstract

Toward Reliable Off Road Autonomous Vehicles Operating in Challenging Environments

Cited by 209 publications

References 60 publications

3D Convolutional Neural Networks for landing zone detection from LiDAR

3D Convolutional Neural Networks for landing zone detection from LiDAR

Mind the gap: detection and traversability analysis of terrain gaps using LIDAR for safe robot navigation

Error modeling and calibration of exteroceptive sensors for accurate mapping applications

Contact Info

Product

Resources

About