Human body model control method, device, electronic device and storage medium

Abstract

The present application proposes a control method, device, electronic equipment, and storage medium of a human body model, wherein the method includes: by acquiring characteristic information used to indicate the state of the human body model, inputting the characteristic information into multiple strategy networks of the walking model to obtain each For the control information output by the strategy network, the value network of the walking model is used to score the control information output by each strategy network. According to the score, the target control information is determined from the control information output by each strategy network, and the human body model is controlled according to the target control information. . Compared with the reinforcement learning algorithm in the prior art, the walking model has the characteristics of high data utilization rate and strong robustness. Therefore, the control information output by each policy network is scored through multiple value networks of the walking model, and then By controlling the human body model according to the target control information determined by the score, the human body model can be controlled more accurately, so that the human body model can keep walking in a normal posture.

Description

technical field [0001] The present application relates to the technical field of natural language processing, and in particular to a control method, device, electronic equipment and storage medium of a human body model. Background technique [0002] With the rapid development of computer technology, the use of computer software to model and simulate human musculoskeletal models has become a mainstream. After years of development, foreign experts and scholars have successively developed simulation software for human motion analysis such as SIMM, AnyBody, LifeMod (seamlessly connected with Adams), OpenSim and MSMS. [0003] In recent years, with the development of reinforcement learning and deep learning, reinforcement learning algorithms based on continuous control have emerged, which train a deep neural network policy model to predict control vectors. For example, the Deep Deterministic Policy Gradient algorithm (Deep Deterministic Policy Gradient, DDPG for short) of the Of...

AI Technical Summary

Problems solved by technology

[0005] For this reason, this application proposes a control method, device, electronic equipment and storage medium of a human body model to solve the technical problems of poor robustness and low sample utilization in the reinforcement learning algorithm for human body model training in the prior art

Images