Learning From Sparse Demonstrations

Jin, Wanxin; Murphey, Todd D.; Kulić, Dana; Ezer, Neta; Mou, Shaoshuai

doi:10.1109/tro.2022.3191592

Cited by 10 publications

(5 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the KMP method, the width of Gaussian for each kernel and the scale factor of the regularization term play essential roles in encoding and generalizing the learned trajectory distribution. For both methods, we are using the open-sourced implementations 3 .…”

Section: B Benchmarks Among Learning-from-demonstration Methodsmentioning

confidence: 99%

“…The goal is to compute a probability distribution of the given demonstrations as a reference to guide the future executions of the robot for a similar task. In this work, the widely-recognized group in robotics, i.e., SE (3), is used to derive the proposed method. And the extension to another Lie group, i.e., Pose Change Group (PCG(3)) [50], is also introduced.…”

Section: Probabilistically-informed Motion Primitivesmentioning

confidence: 99%

“…Different demonstrations might have different speed of execution, even when the trajectories have the same shape [3]. Speed difference among demonstrations may cause the same feature to be parameterized at different time steps.…”

Section: B Temporal Alignment For Demonstrations Using Globally Optim...mentioning

confidence: 99%

“…The latter case is much more challenging and related to a popular field in robot learning, namely Learning-from-Demonstration (LfD) [1] or Programming-by-Demonstration (PbD) [2]. Many works on LfD encode demonstrations as trajectories in Euclidean space [3], [4]. For example, the angles and velocities of all joints are considered as a multi-dimensional state vector; or only the positions of the end effector are modeled.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

PRIMP: PRobabilistically-Informed Motion Primitives for Efficient Affordance Learning From Demonstration

Ruan,

Liu,

Wang

et al. 2024

IEEE Trans. Robot.

View full text Add to dashboard Cite

This paper proposes a learning-from-demonstration (LfD) method using probability densities on the workspaces of robot manipulators. The method, named PRobabilistically-Informed Motion Primitives (PRIMP), learns the probability distribution of the end effector trajectories in the 6D workspace that includes both positions and orientations. It is able to adapt to new situations such as novel via points with uncertainty and a change of viewing frame. The method itself is robot-agnostic, in that the learned distribution can be transferred to another robot with the adaptation to its workspace density. Workspace-STOMP, a new version of the existing STOMP motion planner, is also introduced, which can be used as a post-process to improve the performance of PRIMP and any other reachability-based LfD method. The combination of PRIMP and Workspace-STOMP can further help the robot avoid novel obstacles that are not present during the demonstration process. The proposed methods are evaluated with several sets of benchmark experiments. PRIMP runs more than 5 times faster than existing state-of-the-art methods while generalizing trajectories more than twice as close to both the demonstrations and novel desired poses. They are then combined with our lab's robot imagination method that learns object affordances, illustrating the applicability to learn tool use through physical experiments.

show abstract

Section: B Benchmarks Among Learning-from-demonstration Methodsmentioning

confidence: 99%

Section: Probabilistically-informed Motion Primitivesmentioning

confidence: 99%

Section: B Temporal Alignment For Demonstrations Using Globally Optim...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

PRIMP: PRobabilistically-Informed Motion Primitives for Efficient Affordance Learning From Demonstration

Ruan,

Liu,

Wang

et al. 2024

IEEE Trans. Robot.

View full text Add to dashboard Cite

show abstract

“…) with respect to θ, where w * i (θ) is obtained by minimizing the inner-level objective Ψ i (θ, w) parameterized by θ. Bilevel optimization has recently been used in various applications such as meta-learning [52], reinforcement learning [53], robotics [54], [55], etc.…”

Section: Bilevel Optimizationmentioning

confidence: 99%

Beyond Inverted Pendulums: Task-optimal Simple Models of Legged Locomotion

Chen¹,

Hu²,

Posa³

2023

Preprint

View full text Add to dashboard Cite

Reduced-order models (ROM) are popular in online motion planning due to their simplicity. A good ROM captures the bulk of the full model's dynamics while remaining low dimension. However, planning within the reduced-order space unavoidably constrains the full model, and hence we sacrifice the full potential of the robot. In the community of legged locomotion, this has lead to a search for better model extensions, but many of these extensions require human intuition, and there has not existed a principled way of evaluating the model performance and discovering new models. In this work, we propose a model optimization algorithm that automatically synthesizes reducedorder models, optimal with respect to any user-specified cost function. To demonstrate our work, we optimized models for a bipedal robot Cassie. We show in hardware experiment that the optimal ROM is simple enough for real time planning application and that the real robot achieves higher performance by using the optimal ROM.

show abstract