Anticipate your surroundings: Predictive collision detection between dynamic obstacles and planned robot trajectories on the GPU

Hermann, A. M.; Mauch, Felix; Fischnaller, Klaus; Klemm, Sebastian; Roennau, Arne; Dillmann, Ruediger

doi:10.1109/ecmr.2015.7324047

Cited by 24 publications

(12 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our scene monitoring can be added on top of many motion planning and execution pipelines and runs at 20 Hz on CPU, which is responsive enough for operation on a moving platform. Work by Hermann et al [21] checks swept volumes of trajectories on the GPU with additional predictive tracking of obstacles at 6-8 Hz and with replanning using a library of motion primitives. As a result, they interactively adapt the execution speed or interrupt motion if in collision.…”

Section: Discussionmentioning

confidence: 99%

“…Future work includes continuous adaptation and local replanning of motion trajectories in response to environment changes captured by our scene monitoring, and to incorporate motion flow and predictive tracking similar to [21]. Additionally, we are interested in incorporating a novel improvement of dynamic reachability maps [29] leveraging hierarchies in the kinematic structure to reduce memory requirements by several orders of magnitude allowing complete maps for reachability and motion planning to be stored in memory at runtime.…”

Section: Discussionmentioning

confidence: 99%

“…The continuously integrated discretized occupancy map is the basis for tracking changes. Our scene monitoring and reasoning (shown in Figure 5) hereby works similar to the one described in [21]. Instead of analyzing a sequence of swept volumes, we check the key poses of the trajectory currently being executed directly against the collision map and changed areas upon every map update.…”

Section: B Continuous Scene Monitoringmentioning

confidence: 99%

See 2 more Smart Citations

Robust shared autonomy for mobile manipulation with continuous scene monitoring

Merkt

Yang

Stouraitis

et al. 2017

2017 13th IEEE Conference on Automation Science and Engineering (CASE)

View full text Add to dashboard Cite

This work presents a fully integrated system for reliable grasping and manipulation using dense visual mapping, collision-free motion planning, and shared autonomy. The particular motion sequences are composed automatically based on high-level objectives provided by a human operator, with continuous scene monitoring during execution automatically detecting and adapting to dynamic changes of the environment. The system is able to automatically recover from a variety of disturbances and fall back to the operator if stuck or if it cannot otherwise guarantee safety. Furthermore, the operator can take control at any time and then resume autonomous operation. Our system is flexible to be adapted to new robotic systems, and we demonstrate our work on two real-world platformsfixed and floating base -in shared workspace scenarios.To the best of our knowledge, this work is also the first to employ the inverse Dynamic Reachability Map for realtime, optimized mobile base positioning to maximize workspace manipulability reducing cycle time and increasing planning and autonomy robustness.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: B Continuous Scene Monitoringmentioning

confidence: 99%

See 1 more Smart Citation

Robust shared autonomy for mobile manipulation with continuous scene monitoring

Merkt

Yang

Stouraitis

et al. 2017

2017 13th IEEE Conference on Automation Science and Engineering (CASE)

View full text Add to dashboard Cite

show abstract

“…Doing this for every point however is computationally burdensome, and the algorithm proved to not execute in real time, with a maximum frequency of 6 to 8 Hz (Herman et al, 2015). collision detection algorithm that can operate on much more basic predictions of the human's location and orientation throughout time.…”

Section: Related Workmentioning

confidence: 99%

Human Modeling for Efficient Predictive Collision Detection

Streitmatter

2020

UFJUR

View full text Add to dashboard Cite

As demands on manufacturing rapidly evolve, flexible manufacturing is becoming more essential for acquiring the necessary productivity to remain competitive. An innovative approach to flexible manufacturing is the introduction of fenceless robotic manufacturing cells to acquire and leverage greater human-robot collaboration (HRC). This involves operations in which a human and a robot share a space, complete tasks together, and interact with each other. Such operations, however, pose serious safety concerns. Before HRC can become a viable possibility, robots must be capable of safely operating within and responding to events in dynamic environments. Furthermore, the robot must be able to do this quickly during online operation. This paper outlines an algorithm for predictive collision detection. This algorithm gives the robot the ability to look ahead at its trajectory, and the trajectories of other bodies in its environment and predict potential collisions. The algorithm approximates a continuous swept volume of any articulated body along its trajectory by taking only a few time sequential samples of the predicted orientations of the body and creating surfaces that patch the orientations together with Coons patches. Run time data collected on this algorithm suggest that the algorithm can accurately predict future collisions in under 30 ms.

show abstract

“…Collision checking with the full mesh model of the object is time-consuming. The required time can be reduced by using a GPGPU unit [21], [22] but this type of the hardware is not available on all types of the robot because of the energy consumption and maximum payload of small walking machines. Therefore, the first idea to speed-up collision checking is to simplify the mesh model of the robot.…”

Section: Self-collisionsmentioning

confidence: 99%

Efficient Modeling and Evaluation of Constraints in Path Planning for Multi-Legged Walking Robots

Belter

2019

IEEE Access

View full text Add to dashboard Cite

In this article, we propose a new method for constraints evaluation during the path planning of a multi-legged walking robot. We propose the application of Gaussian Mixtures (GM) to build constraint models. With the proposed analytical constraint function we can compute the gradient and move the robot out of the bound configuration. The proposed method allows checking self-collisions, and a workspace of the robot in a few microseconds and efficiently plan the motion of the robot. The proposed path planning method uses RRT-Connect framework. We show how to efficiently apply the constraints evaluation methods to optimize the posture of the robot, optimize the position of the feet during the swing phase and efficiently explore the search space. We compare the proposed method with the standard constraints evaluation algorithms. Finally, we present the results of path planning for the six-legged robot in various scenarios to show properties of the proposed motion planning algorithm with GM-based constraints evaluation. INDEX TERMS Autonomous systems, legged locomotion, path planning, robot motion.

show abstract

Anticipate your surroundings: Predictive collision detection between dynamic obstacles and planned robot trajectories on the GPU

Cited by 24 publications

References 16 publications

Robust shared autonomy for mobile manipulation with continuous scene monitoring

Robust shared autonomy for mobile manipulation with continuous scene monitoring

Human Modeling for Efficient Predictive Collision Detection

Efficient Modeling and Evaluation of Constraints in Path Planning for Multi-Legged Walking Robots

Contact Info

Product

Resources

About