Posture Risk Assessment and Workload Estimation for Material Handling by Computer Vision

Table 1

Different methods of human posture risk assessment.

Category	Device	Approach	Contribution	References
Contact sensor method	MoCap	Collect joint coordinates	An inertial sensor approach was developed to evaluate the working posture of workers on an automotive assembly line	[13]
	IMU sensor	Collect joint coordinates	A real-time motion alert personal protective equipment (PPE) based on a wearable inertial measurement unit (Wimu) was developed to enable workers to self-awareness and self-management of ergonomic hazardous operating modes to prevent WMSDs	[14]
Noncontact vision method	Kinect v2	Human key point recognition	Proposing a semiautomatic RULA evaluation software based on Microsoft Kinect v2 depth camera	[15]
	Kinect v2	Depth image acquisition and segmentation	A skeleton less holistic pose analysis system is proposed to accurately predict operational pose limb angles from a single depth image	[16]
	RGB-D camera	2D pose estimation based on CEPARA	A PAS-oriented holistic pose acquisition and ergonomic risk analysis model is proposed for developing a smartphone-based and workplace-based WMSDs risk assessment system	[17]
	RGB-D camera	2D pose estimation	A proposed vision-based real-time RULA method for scoring RULA levels of individual images captured by ordinary RGB cameras	[18]
	RGB-D camera	3D pose estimation based on 3DMPPE	The body angle reliability decision making (BARD) method is proposed to calculate the most reliable body bending angle	[2]