CHOMP: Gradient optimization techniques for efficient motion planning
By Nathan D. Ratliff, Matthew Zucker, J. Bagnell, Siddhartha S. Srinivasa. 2009
1,191 citations recorded by the source corpus.
Existing high-dimensional motion planning algorithms are simultaneously overpowered and underpowered. In domains sparsely populated by obstacles, the heuristics used by sampling-based planners to navigate “narrow passages” can be needlessly complex; furthermore, additional post-processing is required to remove the jerky or extraneous motions from the paths that such planners generate. In this paper, we present CHOMP, a novel method for continuous path refinement that uses covariant gradient techniques to improve the quality of sampled trajectories. Our optimization technique both optimizes higher-order dynamics and is able to converge over a wider range of input paths relative to previous path optimization strategies. In particular, we relax the collision-free feasibility prerequisite on input paths required by those strategies. As a result, CHOMP can be used as a standalone motion planner in many real-world planning queries. We demonstrate the effectiveness of our proposed method in manipulation planning for a 6-DOF robotic arm as well as in trajectory generation for a walking quadruped robot.
Subfields: Aerospace Engineering, Artificial Intelligence, Computer Vision and Pattern Recognition, Mathematical Optimization, Robotics & Control
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Aerospace Engineering
- Pieter Abbeel
- Vladlen Koltun
- Sergey Levine
- Shaojie Shen
- Siddhartha S. Srinivasa
- Dhruv Batra
- Jitendra Malik
- Ashutosh Saxena
In Artificial Intelligence
- Sergey Levine
- Pieter Abbeel
- Chelsea Finn
- Yoshua Bengio
- Ruslan Salakhutdinov
- Trevor Darrell
- Mohit Bansal
- Quoc V. Le
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Jitendra Malik
- Trevor Darrell
- Dacheng Tao
- Abhinav Kumar Gupta
- Sergey Levine
- Xiaogang Wang
- Ming-Hsuan Yang
In Mathematical Optimization
- Siddhartha S. Srinivasa
- Anca Dragan
- Peter Pastor
- Mrinal Kalakrishnan
- Stefan Schaal
- Dmitry Berenson
- Pieter Abbeel
- Sachin Chitta
In Robotics & Control
- Sergey Levine
- Pieter Abbeel
- Siddhartha S. Srinivasa
- J. Bagnell
- Thomas Lampe
- Martin A. Riedmiller
- Mrinal Kalakrishnan
- Peter Pastor
Citations in the corpus
- Learning Trajectory Preferences for Manipulators via Iterative Improvement (2013, methodology): “The cost function (or its approximation) we learn can be fed to trajectory optimizers like CHOMP [29] or optimal planners like RRT* [15] to produce reasonably good trajectories. α(s∗(xt, y ∗ t ) − s∗(xt, yt)) − ξt.”
- Habitat 2.0: Training Home Assistants to Rearrange their Habitat (2021, background): “For instance, work in the area of motion planning has made several advancements with new sampling techniques [94, 95] and optimization based methods [86, 87, 89, 96], but largely operated on the assumption of a reliable perception stack.”
- Generating Legible Motion (2013, methodology): “7 an example using the cost functional from the CHOMP motion planner[29], which trades off between the sum-squared velocity cost we have been using thus far, and a cost penalizing the robot from coming too close to obstacles.”
- Voxblox: Incremental 3D Euclidean Signed Distance Fields for on-board MAV planning (2016, background): “However, they require the distances to obstacles to be known at all points in a map, as well as distance gradients [1], [2].”
- Acquiring visual servoing reaching and grasping skills using neural reinforcement learning (2013, methodology): “generation of trajectories using only a generic initializa tion [6], [23].”
- Perceiving, learning, and exploiting object affordances for autonomous pile manipulation (2013, methodology): “We use CHOMP [18] to generate smooth trajectories and rely on a library of force feedback compliant motion primitives that are safe and appropriate for manipulation under uncertainty [19].”
- A policy-blending formalism for shared control (2013, methodology): “Given such a cost, the robot would use a trajectory optimizer (Todorov and Li, 2005; Ratliff et al., 2009; Toussaint, 2009; Dragan et al., 2011b; Kalakrishnan et al., 2011) to acquire the intended trajectory (Figure 4).”
- A robot path planning framework that learns from experience (2012, methodology): “There are several methods in motion planning [3] and trajectory optimization [20] [21] that are capable of this task.”
- Motion planning with sequential convex optimization and convex collision checking (2014, methodology): “TrajOpt and other optimization approaches such as CHOMP [41, 61] can be used for trajectory smoothing in such cases.”
- Vision-Only Robot Navigation in a Neural Radiance World (2021, neutral): “Perhaps closest to this work’s trajectory optimizer is CHOMP [25], [26], a family of gradient-based methods which”
- STOMP: Stochastic trajectory optimization for motion planning (2011, methodology): “Multiplication of the cost by the magnitude of the workspace velocity of the sphere (‖ẋb‖) avoids this effect [9].”
- A unified framework for coordinated multi-arm motion planning (2018, background): “As the distance from any point in a sphere to the nearest obstacle is lower-bounded to d ( c ) − r , where c is the center of the sphere and r its corresponding radius (Ratliff et al. 2009).”
- Finding Locally Optimal, Collision-Free Trajectories with Sequential Convex Optimization (2013, background): “CHOMP (Covariant Hamiltonian Optimization for Motion Planning) is a set of ideas for how to formulate the objective in robotic motion planning problems and how to perform the numerical optimization [25, 32, 9].”
- Human-robot collaborative manipulation planning using early prediction of human motion (2013, methodology): “We rely on recent developments in trajectory optimization for motion planning [1] [22] [23] to compute low-cost robot motion plans quickly.”
- Chisel: Real Time Large Scale 3D Reconstruction Onboard a Mobile Device using Spatially Hashed Signed Distance Fields (2015, methodology): “In robotics, distance fields are used in motion planning [25], mapping [26], and scene understanding.”
- Learning preferences for manipulation tasks from online coactive feedback (2015, methodology): “Recent trajectory optimizers such as CHOMP [47] and TrajOpt [50] provide optimization based approaches to finding optimal trajectory.”