Static Force and Position Feedforward Control of an Antagonistic McKibben Joint Based on Experimentally Identified Inverse Maps
Year: 2026
Type: article
Abstract: Pneumatic artificial muscles (PAMs) combine high force-to-mass ratio and intrinsic compliance, but their nonlinear and hysteretic force–pressure–contraction behavior complicates control. This study investigates a static feedforward strategy for a one-degree-of-freedom joint actuated by two custom McKibben muscles in an agonist–antagonist arrangement. Each actuator was experimentally characterized under isometric conditions over contraction ratios from 0% to 35% and pressure up to 2.75 bar. The resulting static surfaces were inverted to determine the two pressure command maps required for a prescribed joint angle, co-contraction force, and signed tangential output-force demand. A preliminary MATLAB/Simulink (R2017b) model was used to verify the feasibility of the two-map command concept before hardware implementation. Validation was then performed on a dedicated sprocket-chain test bench. Static positioning tests at co-contraction levels from 150 to 400 N produced mean absolute angular errors from 2.0° to 10.7°, depending on the operating region. Output-force tests were conducted at 0–50° in both clockwise and counter-clockwise directions; across 223 command points, the median absolute percentage error (MdAPE) was 15.3%. Favorable operation was observed in the CCW direction at 30–50° with F0 = 100 N, where MdAPE was approximately 3–4%, whereas high co-contraction (F0 = 300 N) produced markedly poorer performance, with MdAPE values of 59.3% over 0–10° CW and 84.0% over 0–20° CCW. The maximum tested output-force command was 1250 N at 50°, where the measured antagonistic force difference reached 1138 N. The results show that experimentally identified inverse maps can provide simple and directly interpretable static force–position commands, provided that the configuration-dependent operating envelope is explicitly considered.
Source: Machines
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,896
OpenAlex ID: W7221552575