Neuromorphic Perception for Greenhouse Technology
Research Student
Research Project Details
The past 60 years of research have been devoted to frame-based cameras, however, they are still not good enough as they provide standard stroboscopic synchronous sequences of 2D pictures. They still suffer from relatively high latency, motion blur and low dynamic range. Event-based cameras are novel neuromorphic sensors which do not suffer from all these problems. The output of event-based cameras consists of a sequence of digital spikes with each event representing a change in brightness in log intensity and capable of significantly reducing the amount of data required for processing.
My research seeks to leverage the technological advantages of the event camera and investigate their capabilities as a low power device to enable rapid object tracking and multi-class identification using spike-based learning algorithm. My research will also enhance the vision capability of these biologically inspired devices by introducing an asynchronous time-based image sensor that is sensitive to colour information.
The outcome of this research is to use these sensors in a real-world environment to perform automatic and less human involved fruit picking, therefore, the work will be tested and evaluated at the Hawkesbury Institute of the Environment's greenhouse.
