Masters student from University of Moratuwa. Engaging in haptic based object classification. My supervisor is Dr A.M. Harsha.S. Abeykoon. I am willing to work until robots extrapolate the market.

Introduction

Vision-based approaches are extensively employed in research to classify objects. However, there are specific characteristics of deformable objects that are challenging to classify using vision-based sensors. Examples include vegetables that may seem fresh externally but are rotten internally, ripe or unripe fruits, and objects designed to mimic real-life objects. In such cases, the texture or feel of the object may reveal distinct characteristics or its state, which may facilitate classification.

Unfortunately, the majority of object classification studies rely solely on sensory information for classification. The objective of my research is to introduce additional features for object classification and identify the optimal performing feature set. Furthermore, we aim to use time series forest classifier machine learning models to classify the objects.

Additional reference

The summary of the project can be found here.

Data acquisition video footage

Force controller and data acquisition. GitHub

Data anlysis and feature creation. GitLab