Ground reaction forces (GRFs) on individual support vary with posture and motion for bipedal mechanisms or systems due to the redundancy in the system. In digital human modeling, specifically posture prediction, the GRFs are predicted, as they are unknown in a virtual environment. Traditionally, models in which the GRFs are predicted have been presented; however, they are always assumed to be on flat ground. Little work has been done to predict the GRFs on uneven or arbitrary terrain. This paper presents a generic method to calculate the vertical GRFs for given standing postures with uneven terrain. The vertical GRFs are predicted based on the generalized forces (torque in revolute joints; force in prismatic joints) calculated using the recursive Lagrangian formulation and a 3-D zero moment point. Motion capture experiments were used to obtain postures for common standing reaching tasks. Force plates were employed to record GRF information for each task. Experimental postures were reconstructed, and the GRF prediction algorithm was used to predict the associated vertical GRFs for each task. Experimental and predicted vertical GRFs are compared to validate the prediction model. The prediction method proved to be valid, with an overall error of 6%.
- Digital human model
- Ground reaction forces (GRFs)
- Standing posture
- Uneven terrain