A Low-Cost Real-Time Spiking System for Obstacle Detection based on Ultrasonic Sensors and Rate Coding

TL;DR

An in-depth study on how this system works at low level based on the Inter-Spike Interval concept was performed, which may be useful in the future development of applications based on spiking filters.

Abstract

Since the advent of mobile robots, obstacle detection has been a topic of great interest. It has also been a subject of study in neuroscience, where flying insects and bats could be considered two of the most interesting cases in terms of vision-based and sound-based mechanisms for obstacle detection, respectively. Currently, many studies focus on vision-based obstacle detection, but not many can be found regarding sound-based obstacle detection. This work focuses on the latter approach, which also makes use of a Spiking Neural Network to exploit the advantages of these architectures and achieve an approach closer to biology. The complete system was tested through a series of experiments that confirm the validity of the spiking architecture for obstacle detection. It is empirically demonstrated that, when the distance between the robot and the obstacle decreases, the output firing rate of the system increases in response as expected, and vice versa. Therefore, there is a direct relation between the two. Furthermore, there is a distance threshold between detectable and undetectable objects which is also empirically measured in this work. An in-depth study on how this system works at low level based on the Inter-Spike Interval concept was performed, which may be useful in the future development of applications based on spiking filters.

Figures (8)

• Figure 1: General diagram of the complete system. The three main blocks of which it is composed are shown: a robotic platform, the SpiNNaker neuromorphic platform and a computer, together with the type of information that is transmitted between them and the protocols used.
• Figure 2: Picture of the robotic platform showing its different components: A) Romeo BLE board. B) Adafruit HUZZAH32. C) HC-SR04 ultrasonic sensor.
• Figure 3: General diagram of the implemented SNN.
• Figure 4: Example of output neuron response to an input spike train. At the top of the figure, the membrane potential calculated using the equations presented in rhodes2018spynnaker and the membrane potential obtained from the SpiNNaker hardware platforms after simulation are compared. The upper black line delimits the threshold potential, while the lower black line delimits the minimum potential necessary to fire an output spike when an input spike is received. At the middle, the synaptic currents are shown. At the bottom, a comparison of the ISI of input spikes and the firing windows is made. Input spikes are marked with magenta crosses if they do not cause the output neuron to fire, or with red crosses if they do.
• Figure 5: System response to measurements sent by the Romeo BLE board corresponding to an object located at a constant distance of approximately 39 cm (top) and 39.5 cm (bottom). Output spikes are marked with red points and vertical lines.