Let’s start with what is Neuromorphic Computing and why all of a sudden researchers are getting attracted to it? So far, many computing techniques are implemented for different general/ specific applications, but energy efficiency is still a major issue in every computation. Especially in the case of training a neural network where data might contain millions of features, a huge amount of energy is needed for computation. With a boom in artificial intelligence based applications, researchers are focusing to develop a system that can perform complex computations with less energy consumption.
Many researchers in past tried to draw motivation from how the human brain performs. Efforts were made to understand the complexity of the human brain and how it computes so much information with so less energy in a short amount of time. The human brain uses only 20 watts and this amount of energy it can do wonders, which we see in daily life. Current computer technology based on Von-Neumann architecture is still far behind. Though the recent development in the area of AI opens many doors to make computers perform tasks like multi-classification, segmentation, natural language processing, etc, there is still a lot to do in this area.
Neuromorphic computing is a method of computer engineering in which elements of a computer are modeled after the system in the human brain. Unlike continuous computation in artificial neural networks, spike-driven computation is used where neurons are actual nodes and synapses are the connections between those nodes. The network of neurons and synapses that transmit spikes (electric pulses) is known as the Spiking Neural Network (SNN).
The interaction between humans and machines is of great relevance for both the field of neuromorphic computing and Robotics. Utilizing neuromorphic technologies in robotics, from perception to motor control, is a promising approach to creating robots that can seamlessly integrate into society. In neurorobotics (neuromorphic computing and robotics), bio-inspired sensors are used to efficiently encode sensory signals. It also adapts to different environmental conditions by integrating inputs from multiple sensors and using event-based computation to accomplish the desired task. The use of SNNs in robotics introduces considerable complexity with limited benefits when performing simple tasks. In cognitive robotics, the goal is to understand the environment and compute the output. Such an approach usually returns useful insights for neural architectures and learned behavior, especially when dedicated hardware is available.
In our latest review of papers related to neuromorphic computing for interactive robotics few experiments show the social interaction between robots and humans in real-world scenarios. Even in these experiments, robots have several limitations. Besides this, the lack of universal training methods and conversion mechanisms is also a major challenge in the development of neurorobotics for social interactions. Therefore, more collaborations, especially between roboticists and neuroscientists, are needed to explore this area of neuromorphic computing and social robotics in the future.