Ballistic landers enable orbiting asteroid missions to perform surface science at limited additional cost and risk. Due to asteroids’ weak gravity and irregular terrain, lander deployment trajectories will consist of several chaotic bounces. Although impacts on regolith-covered asteroids are numerically expensive to model, impacts on rocky asteroids can be modeled with simpler, impulsive contact models. One such model is that by Stronge, which was successfully used in large-scale Monte Carlo studies of asteroid lander deployment. This model parameterizes impacts with (fixed) material restitution and friction coefficients, but has not been validated for the low-velocity regime of an assembled, nonspherical body. This paper uses an air-bearing setup to perform 2D experiments of a rectangular floating assembly impacting a concrete block with V⩽25 cm/s. The impact velocity, assembly attitude, and block attitude are varied across 2,400 experimental runs of both normal and tangential impacts. Optical tracking is used to extract the pre- and post-impact velocities of the assembly. In a majority of cases, Stronge's model can be fit to the experiments to extract the corresponding restitution and friction coefficients. We find that the coefficients are not fixed with respect to the impact velocity and attitude, but that their variation is seemingly random. In some tangential impact cases, the model even fails to reproduce the observed behavior althogether. This suggests that there may not be a simple way to reconcile Stronge's fixed-material-coefficient model with reality, although it may retain practical use if the coefficients are randomly varied in each impact of a simulation.