Human Motion Prediction using Semi-adaptable Neural Networks

Cheng, Yujiao; Zhao, Weiye; Liu, Changliu; Tomizuka, Masayoshi

doi:10.48550/arxiv.1810.00781

Cited by 1 publication

(5 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the formulation in [6], the transition model of human motion on a selected joint can be defined as:…”

Section: Human Motion Prediction Methodsmentioning

confidence: 99%

“…RNNs receives increasing interests in human motion prediction recently. However, RNNs is hard to train, many attempts to accelerate the speed have failed [6]. Moreover, since RNNs models are fixed, their recurrent network architectures are not compatible with online adaptation schemes.…”

Section: B Neural Network Model Without Adaptationmentioning

confidence: 99%

“…The last-layer weight W in the feed-forward neural network ( 5) is also adaptable online using RLS-PAA [6]. We call the new model the semi-adaptable neural network.…”

Section: Neural Network Model With Adaptationmentioning

confidence: 99%

“…θ k denotes the value of the parameter vector at time step k. To represent the extracted features, we define a new data matrix Φ k ∈ R 3M×3Mn h as a diagonal concatenation of M pieces of max(0, g(U, s k )). Using Φ k and θ k , then (1) and ( 5) can be written into the same form as (6). The procedure of semi-adaptable neural network is summarized in Algorithm 1.…”

Section: Neural Network Model With Adaptationmentioning

confidence: 99%

“…Moreover, many challenging HRC tasks, such as aircraft and electronics assembly [5], require human workers to work in close proximity to robots. To enable safe and efficient HRC, it is crucial for robots to be aware of current and future human movements, so that robots can quickly adapt their behavior to better collaborate with human workers [6]. However, it is inherently difficult to predict human motion, since human motion is stochastic in nature and subjected to nonlinear dynamics [7].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Experimental Evaluation of Human Motion Prediction: Toward Safe and Efficient Human Robot Collaboration

Zhao

Sun

Liu

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Human motion prediction is non-trivial in modern industrial settings. Accurate prediction of human motion can not only improve efficiency in human robot collaboration, but also enhance human safety in close proximity to robots. Among existing prediction models, the parameterization and identification methods of those models vary. It remains unclear what is the necessary parameterization of a prediction model, whether online adaptation of the model is necessary, and whether prediction can help improve safety and efficiency during human robot collaboration. These problems result from the difficulty to quantitatively evaluate various prediction models in a closedloop fashion in real human-robot interaction settings. This paper develops a method to evaluate the closed-loop performance of different prediction models. In particular, we compare models with different parameterizations and models with or without online parameter adaptation. Extensive experiments were conducted on a human robot collaboration platform. The experimental results demonstrated that human motion prediction significantly enhanced the collaboration efficiency and human safety. Adaptable prediction models that were parameterized by neural networks achieved the best performance.

show abstract

“…Following the formulation in [6], the transition model of human motion on a selected joint can be defined as:…”

Section: Human Motion Prediction Methodsmentioning

confidence: 99%

Section: B Neural Network Model Without Adaptationmentioning

confidence: 99%

“…The last-layer weight W in the feed-forward neural network ( 5) is also adaptable online using RLS-PAA [6]. We call the new model the semi-adaptable neural network.…”

Section: Neural Network Model With Adaptationmentioning

confidence: 99%

Section: Neural Network Model With Adaptationmentioning

confidence: 99%