Abstract:The understanding of human intent based on human motions remains a highly relevant and challenging research topic. The relationship of the sequence of human motions may be a possible solution to recognize human intention. The supervised multiple timescale recurrent neural network (supervised MTRNN) model is a useful tool for motion classification. In this paper, we propose a new model to understand human intention based on human motions in real-time through a deep structure including two supervised MTRNN models, which are based on understanding the meaning of a series of human motions. The 1st supervised MTRNN layer classifies motion labels while the 2nd supervised MTRNN layer in the deep dynamic neural structure identifies human intention using the results of the 1st supervised MTRNN. We also considered the action-perception cycle effect between the 1st and the 2nd supervised MTRNNs, in which the motion label perception and internal action (motion prediction) form a cycle to improve the motion classification and intent recognition performance. A group of tasks was designed around movements involving two objects in an attempt to detect different motions and intentions based on the proposed deep dynamic neural model. The experimental results showed the deep supervised MTRNN to be more robust and to outperform the single layer supervised MTRNN model for detecting human intention. The action-perception cycle was found to efficiently improve both motion classification and prediction, which is important for human intent recognition. We developed an online deep dynamic neural model for intention classification.We evaluated the importance of internal action generation in motion classification and intention classification.Our proposed model performances better than a single layer supervised MTRNN.The possibility of each intention is able to be detected based on our model.

Multiple Timescale Recurrent Neural Network with Slow Feature Analysis for Efficient Motion Recognition.

Continuous Motion Recognition Using Multiple Time Constant Recurrent Neural Network With A Deep Network Model

Real-time human action classification using a dynamic neural model.

Supervised Multiple Timescale Recurrent Neuron Network Model for Human Action Classification.

A Fast Training Algorithm of Multiple-Timescale Recurrent Neural Network for Agent Motion Generation

Recognition of Visually Perceived Compositional Human Actions by Multiple Spatio-Temporal Scales Recurrent Neural Networks

Continuous Timescale Long-Short Term Memory Neural Network for Human Intent Understanding

Human motion based intent recognition using a deep dynamic neural model

Predictive Coding for Dynamic Vision : Development of Functional Hierarchy in a Multiple Spatio-Temporal Scales RNN Model

Multi‐mode Neural Network for Human Action Recognition

Motion Sequence Recognition with Multi-sensors Using Deep Convolutional Neural Network.

Markov Blanket Based Sequential Data Feature Selection for Human Motion Recognition.

DMS-GCN: Dynamic Mutiscale Spatiotemporal Graph Convolutional Networks for Human Motion Prediction

Human Action Recognition by Fast Dense Trajectories.

Human Intention Understanding Based On Object Affordance And Action Classification

Multi-Level Recurrent Residual Networks for Action Recognition

Human Activity Recognition Based on Deep-Temporal Learning Using Convolution Neural Networks Features and Bidirectional Gated Recurrent Unit With Features Selection

MSTP-Net: Multiscale Spatio-temporal Parallel Networks for Human Motion Prediction

Motion-Pose Recurrent Neural Network With Instantaneous Kinematic Descriptor For Skeleton Based Gesture Detection And Recognition

Dynamic Action Recognition: A convolutional neural network model for temporally organized joint location data

3D-Cnn-based Fused Feature Maps with LSTM Applied to Action Recognition.