AirCapRL: Autonomous Aerial Human Motion Capture Using Deep Reinforcement Learning

Abstract

In this letter, we introduce a deep reinforcement learning (DRL) based multi-robot formation controller for the task of autonomous aerial human motion capture (MoCap). We focus on vision-based MoCap, where the objective is to estimate the trajectory of body pose and shape of a single moving person using multiple micro aerial vehicles. State-of-the-art solutions to this problem are based on classical control methods, which depend on hand-crafted system and observation models. Such models are difficult to derive and generalize across different systems. Moreover, the non-linearities and non-convexities of these models lead to sub-optimal controls. In our work, we formulate this problem as a sequential decision making task to achieve the vision-based motion capture objectives, and solve it using a deep neural network-based RL method. We leverage proximal policy optimization (PPO) to train a stochastic decentralized control policy for formation control. The neural network is trained in a parallelized setup in synthetic environments. We performed extensive simulation experiments to validate our approach. Finally, real-robot experiments demonstrate that our policies generalize to real world conditions.

RETRACTED ARTICLE: Human motion capture system in sports performance based on Internet of Things technology and wireless inertial sensor

“…Understanding the mathematical model of the state vector and the observation vector, the characteristics of the statistical noise of the state and the means of observation and the initial value of the state of the system, the measured data and the sensor state equation can be used to derive the relationship between the system state medium and the observation data. Kalman filtering is divided into two stages: prediction and information [9,10]. In the prediction phase, the state estimation at each moment is estimated based on the previous state value.…”

3D Human Reconstruction in the Wild with Collaborative Aerial Cameras

“…Subsequent works investigate mobile UAVs as an alternative to overcome the complexity of this static setup, with [33]- [35] focusing on optimal camera plans for a single UAV. We find works that plan optimal viewpoints with multiple vehicles [21], [36]; however, they do not consider obstacle avoidance. Most related to our work, [23] introduces a decentralized multi-UAV coordination framework for actor position estimation that is extended as a data collection system for outdoor human shape estimation [22].…”

“…For instance, [20] uses multiple drones to capture the pose of a human wearing markers. [21] explicitly optimizes for body pose reconstruction, but in obstacle-free environments. Most related to our work, [22] presents an aerial motion capture system that uses multi-robot formation controller [23] for data collection.…”

Simulation of Gymnastics Performance Based on MEMS Sensor

“…Aerobics, cheerleading, sports dance, group gymnastics, recreational gymnastics, rhythmic gymnastics and other items of technical movements, costumes, props, music and other items constitute the main elements of performing gymnastics. The understanding of performing gymnastics in this study is to weaken its competitive nature, pay more attention to its performance, and make it more visible and entertaining [21][22].…”