Journal: Proceedings. Florida Conference on Recent Advances in Robotics

Article Title: Flexible Magnetic Skin Sensor Array for Torsion Perception

doi: 10.5038/swid5066

Figure Lengend Snippet: ANN training accuracy confusion matrix generated using the MATLAB (MathWorks) neural network pattern recognition toolbox (nprtool). This was taken from one of the 10 networks that were trained. Output classes represent the predicted classifications made by the network, 1 through 10 being the degree variations, 11 being normal force data, and 12 being no touch data. The Target classes on the bottom axis represent the correct classifications for the data.

Article Snippet: The ANN was trained using MATLAB’s Neural Net Pattern Recognition toolbox (nprtool), and the network parameters were modified according to the error generated with the training dataset.

Techniques: Generated

Journal: IEEE Haptics Symposium : [proceedings]. IEEE Haptics Symposium

Article Title: Robotically Embodied Biological Neural Networks to Investigate Haptic Restoration with Neuroprosthetic Hands

doi: 10.1109/haptics52432.2022.9765605

Figure Lengend Snippet: Resulting flow of data from the finger tapping experiments for real-time closed-loop experiments and offline analysis. For the online flow, (c) shows an extracellular recording, (d) shows the event detection based on defined thresholds, (e) is the MEAout signal controlling the finger joint angle θ. (f), (g) contains the FDC and FAC values from the BioTac and (h) shows an example of the SA pulse trains generated by the Izhikevich model (5), (7)-(9) for the SA tactile events. (i,j,k,l) show a zoomed window of the data (a,b,c,d). For the offline flow, part (a) shows a frequency analysis of the spikes extracted in (d) and part (b) shows a frequency analysis of the raw electrode recordings from (c). (m,n) show two different cultures’ classification results for the two time windows of 1s and 2s that were tested with the pattern recognition neural network to distinguish between SA and RA stimulation patterns. (m) shows classifications accuracies for a culture 1 and (n) Culture 2, with each color correlated to the data in (a,b,c,d). (o) shows the comparison of between ITIs produced by the RA and SA encoding methods over the course of 4 days. (*) and (**) signify the statistical significances of p<0.05 and p<0.01 respectively.

Article Snippet: After conducting the online experiments, the recorded BNN data was classified offline via neural network pattern recognition (nprtool, MATLAB).

Techniques: Generated, Comparison, Produced

Journal: Proceedings. Florida Conference on Recent Advances in Robotics

Article Title: Flexible Magnetic Skin Sensor Array for Torsion Perception

doi: 10.5038/swid5066

Figure Lengend Snippet: ANN training accuracy confusion matrix generated using the MATLAB (MathWorks) neural network pattern recognition toolbox (nprtool). This was taken from one of the 10 networks that were trained. Output classes represent the predicted classifications made by the network, 1 through 10 being the degree variations, 11 being normal force data, and 12 being no touch data. The Target classes on the bottom axis represent the correct classifications for the data.

Article Snippet: ANN training accuracy confusion matrix generated using the MATLAB (MathWorks) neural network pattern recognition toolbox (nprtool).

Techniques: Generated

Journal: Sensors (Basel, Switzerland)

Article Title: Robotic Replica of a Human Spine Uses Soft Magnetic Sensor Array to Forecast Intervertebral Loads and Posture after Surgery

doi: 10.3390/s22010212

Figure Lengend Snippet: Robotically probing soft magnet to classify location and amplitude of external loads. ( ai ) The stretchable soft magnet. ( aii ) The Hall effect sensor array PCB with dimensional units of mm. ( aiii ) Each individual Hall effect sensor taxel position was numbered and physical dimensions were labeled with units of mm. ( b ) The UR5 applied loads to the soft magnet that was placed atop the 3 × 3 Hall effect sensor array within the 3D-printed housing. These nine Hall effect sensor signals were recorded in Simulink and classified with four machine learning algorithms in MATLAB.

Article Snippet: The neural network toolbox in MATLAB (nprtool) was used to generate a feedforward network to train and test the ANN [ ].

Techniques: Labeling