Provably safe robot navigation with obstacle uncertainty

Axelrod, Brian; Kaelbling, Leslie Pack; Lozano-Pérez, Tomás

doi:10.1177/0278364918778338

Cited by 55 publications

(40 citation statements)

References 23 publications

(33 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our problem it is the environment that is uncertain rather than the dynamics. One approach to this problem involves casting "shadows" around obstacles [10] but that does not facilitate the resolution of uncertainty from multiple different sources. Probability density functions can usefully model robot and obstacle uncertainty, as in [11], which however requires Monte-Carlo computation and has O(N K) complexity.…”

Section: B Existing Approachesmentioning

confidence: 99%

See 1 more Smart Citation

FPR—Fast Path Risk Algorithm to Evaluate Collision Probability

Blake¹,

Bordallo²,

Brestnichki³

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

As mobile robots and autonomous vehicles become increasingly prevalent in human-centred environments, there is a need to control the risk of collision. Perceptual modules, for example machine vision, provide uncertain estimates of object location. In that context, the frequently made assumption of an exactly known free-space is invalid. Clearly, no paths can be guaranteed to be collision free. Instead, it is necessary to compute the probabilistic risk of collision on any proposed path.The FPR algorithm, proposed here, efficiently calculates an upper bound on the risk of collision for a robot moving on the plane. That computation orders candidate trajectories according to (the bound on) their degree of risk. Then paths within a user-defined threshold of primary risk could be selected according to secondary criteria such as comfort and efficiency.The key contribution of this paper is the FPR algorithm and its 'convolution trick' to factor the integrals used to bound the risk of collision. As a consequence of the convolution trick, given K obstacles and N candidate paths, the computational load is reduced from the naive O(N K), to the qualitatively faster O(N + K).

show abstract

Section: B Existing Approachesmentioning

confidence: 99%

“…As with (10), the right hand side of the inequality (12) is everywhere positive. So, now the mutually complementary expressions (10) and (12) can be combined into a single bound on the indicator function of the Minkowski sum:…”

Section: E Combined Bound For the Fpr Algorithmmentioning

confidence: 99%

FPR—Fast Path Risk Algorithm to Evaluate Collision Probability

Blake¹,

Bordallo²,

Brestnichki³

et al. 2020

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

show abstract

“…To achieve this task robots often rely on raw 2D laser data, despite laser sensors inherent limitation in providing correct distance estimates for many complex objects, including tables, chairs, windows, and open shelfs. A recent paper [12] presented a method for planning theoretically safe paths in environments sensed by a 2D lidar. The authors proposed to inflate obstacles by a volume that represented both the pose uncertainty and the space the obstacle may occupy.…”

Section: Related Work a Uncertainty In Robotic Mobilitymentioning

confidence: 99%

Deep Network Uncertainty Maps for Indoor Navigation

Verdoja

Lundell

Kyrki

2019

2019 IEEE-RAS 19th International Conference on Humanoid Robots (Humanoids)

View full text Add to dashboard Cite

Most mobile robots for indoor use rely on 2D laser scanners for localization, mapping and navigation. These sensors, however, cannot detect transparent surfaces or measure the full occupancy of complex objects such as tables. Deep Neural Networks have recently been proposed to overcome this limitation by learning to estimate object occupancy. These estimates are nevertheless subject to uncertainty, making the evaluation of their confidence an important issue for these measures to be useful for autonomous navigation and mapping. In this work we approach the problem from two sides. First we discuss uncertainty estimation in deep models, proposing a solution based on a fully convolutional neural network. The proposed architecture is not restricted by the assumption that the uncertainty follows a Gaussian model, as in the case of many popular solutions for deep model uncertainty estimation, such as Monte-Carlo Dropout. We present results showing that uncertainty over obstacle distances is actually better modeled with a Laplace distribution. Then, we propose a novel approach to build maps based on Deep Neural Network uncertainty models. In particular, we present an algorithm to build a map that includes information over obstacle distance estimates while taking into account the level of uncertainty in each estimate. We show how the constructed map can be used to increase global navigation safety by planning trajectories which avoid areas of high uncertainty, enabling higher autonomy for mobile robots in indoor settings.

show abstract

“…Finally, a framework has been proposed (Axelrod et al, 2018) to compute shadows as the geometric equivalent of a confidence interval around observed geometric objects which introduces tighter bounds than those of previous methods and the tightness of the bounds does not depend on the number of obstacles by relying on computing a bound specific to a trajectory instead of trying to identify a generic safe set.…”

Section: Related Workmentioning

confidence: 99%

Multiquery Motion Planning in Uncertain Spaces: Incremental Adaptive Randomized Roadmaps

Khaksar

Uddin

Tørresen

2019

International Journal of Applied Mathematics and Computer Science

View full text Add to dashboard Cite

Sampling-based motion planning is a powerful tool in solving the motion planning problem for a variety of different robotic platforms. As its application domains grow, more complicated planning problems arise that challenge the functionality of these planners. One of the main challenges in the implementation of a sampling-based planner is its weak performance when reacting to uncertainty in robot motion, obstacles motion, and sensing noise. In this paper, a multi-query sampling-based planner is presented based on the optimal probabilistic roadmaps algorithm that employs a hybrid sample classification and graph adjustment strategy to handle diverse types of planning uncertainty such as sensing noise, unknown static and dynamic obstacles and an inaccurate environment map in a discrete-time system. The proposed method starts by storing the collision-free generated samples in a matrix-grid structure. Using the resulting grid structure makes it computationally cheap to search and find samples in a specific region. As soon as the robot senses an obstacle during the execution of the initial plan, the occupied grid cells are detected, relevant samples are selected, and in-collision vertices are removed within the vision range of the robot. Furthermore, a second layer of nodes connected to the current direct neighbors are checked against collision, which gives the planner more time to react to uncertainty before getting too close to an obstacle. The simulation results for problems with various sources of uncertainty show a significant improvement compared with similar algorithms in terms of the failure rate, the processing time and the minimum distance from obstacles. The planner is also successfully implemented and tested on a TurtleBot in four different scenarios with uncertainty.

show abstract

Provably safe robot navigation with obstacle uncertainty

Cited by 55 publications

References 23 publications

FPR—Fast Path Risk Algorithm to Evaluate Collision Probability

FPR—Fast Path Risk Algorithm to Evaluate Collision Probability

Deep Network Uncertainty Maps for Indoor Navigation

Multiquery Motion Planning in Uncertain Spaces: Incremental Adaptive Randomized Roadmaps

Contact Info

Product

Resources

About