An approach to validation of a leg simulation by the comparison of two dynamic models

The dynamics of a jointed leg were simulated using two different models, one based on a recursive Newton-Euler method and one on a closed-form Lagrange method. To validate the models, the simulations were run in parallel and the intermediate steps and output of the two methods were compared to one another to reveal the presence and locations of errors. Sources of error and the use of this method for the detection of errors are discussed. Some errors could not have been detected using only a single simulation.