Search for: All records

Objective: Despite advances in human-machine interface design, we lack the ability to give people precise and fast control over high degree of freedom (DOF) systems, like robotic limbs. Attempts to improve control often focus on the static map that links user input to device commands; hypothesizing that the user’s skill acquisition can be improved by finding an intuitive map. Here we investigate what map features affect skill acquisition. Methods: Each of our 36 participants used one of three maps that translated their 19-dimensional finger movement into the 5 robot joints and used the robot to pick up and move objects. The maps were each constructed to maximize a different control principle to reveal what features are most critical for user performance. 1) Principal Components Analysis to maximize the linear capture of finger variance, 2) our novel Egalitarian Principal Components Analysis to maximize the equality of variance captured by each component and 3) a Nonlinear Autoencoder to achieve both high variance capture and less biased variance allocation across latent dimensions Results: Despite large differences in the mapping structures there were no significant differences in group performance. Conclusion: Participants’ natural aptitude had a far greater effect on performance than the map. Significance: Robot-user interfaces are becoming increasingly common and require new designs to make them easier to operate. Here we show that optimizing the map may not be the appropriate target to improve operator skill. Therefore, further efforts should focus on other aspects of the robot-user-interface such as feedback or learning environment. 

This paper proposes a method to learn ap- proximations of missing Ordinary Differential Equations (ODEs) and states in physiological models where knowl- edge of the system’s relevant states and dynamics is in- complete. The proposed method augments known ODEs with neural networks (NN), then trains the hybrid ODE-NN model on a subset of available physiological measurements (i.e., states) to learn the NN parameters that approximate the unknown ODEs. Thus, this method can model an ap- proximation of the original partially specified system sub- ject to the constraints of known biophysics. This method also addresses the challenge of jointly estimating physio- logical states, NN parameters, and unknown initial condi- tions during training using recursive Bayesian estimation. We validate this method using two simulated physiolog- ical systems, where subsets of the ODEs are assumed to be unknown during the training and test processes. The proposed method almost perfectly tracks the ground truth in the case of a single missing ODE and state and performs well in other cases where more ODEs and states are missing. This performance is robust to input signal per- turbations and noisy measurements. A critical advantage of the proposed hybrid methodology over purely data-driven methods is the incorporation of the ODE structure in the model, which allows one to infer unobserved physiological states. The ability to flexibly approximate missing or inac- curate components in ODE models improves a significant modeling bottleneck without sacrificing interpretability.