Casually finishing the day off with us
Major events calendar 2023 - London Convention Bureau.
Read moreThe design of robots that interact autonomously with the environment and exhibit complex behaviors is an open challenge that can benefit from understanding what makes living beings fit to act in the world. Neuromorphic engineering studies neural computational principles to develop technologies that can provide a computing substrate for building compact and low-power processing systems. In this workshop we aim to discuss why endowing robots with neuromorphic technologies – from perception to motor control – represents a promising approach for the creation of robots which can seamlessly integrate in society. Highlighting open challenges in this direction, we propose community participations and actions required to overcome current limitations.
Prof. Jorge Dias has a Ph.D. on EE and Coordinates the Artificial Perception Group from the Institute of Systems and Robotics from the University of Coimbra, Portugal. He is Full Professor at Khalifa University, Abu Dhabi, UAE and Deputy Director from the Center of Autonomous Robotic Systems from Khalifa University. His expertise is in the area of Artificial Perception (Computer Vision and Robotic Vision) and has contributions on the field since 1984. He has been principal investigator and consortia coordinator from several research international projects, and coordinates the research group on Computer Vision and Artificial Perception from KUCARS. Jorge Dias published several articles in the area of Computer Vision and Robotics that include more than 300 publications in international journals and conference proceedings and recently published book on Probabilistic Robot Perception that addresses the use of statistical modeling and Artificial Intelligence for Perception, Planning and Decision in Robots. He was the Project Coordinator of two European Consortium for the Projects “Social Robot” and “GrowMeUP” that were developed to support the inclusivity and wellbeing for of the Elderly generation.
In recent years, there has been a growing interest in developing robotic systems with enhanced perceptual capabilities, enabling them to interact intelligently with their environments. Traditional approaches to robotic perception often rely on algorithms that process sensory information in a step-by-step manner, which can be computationally expensive and lack the flexibility exhibited by biological systems. To address these limitations, the emerging field of neuromorphic engineering offers a promising alternative by drawing inspiration from the structure and function of the human brain. This abstract presents an investigation into the utilization of neuromorphic concepts for robotic perception, aiming to bridge the gap between artificial intelligence and embodied cognition. By leveraging the inherent parallelism and efficiency of neuromorphic hardware, we aim to develop robotic systems that can perceive and understand their surroundings in a manner closer to biological organisms. Our research focuses on the integration of neuromorphic hardware with state-of-the-art machine learning algorithms, allowing robots to process sensory data in real-time and extract meaningful information with minimal computational resources. We propose a hybrid approach that combines the strengths of traditional computer vision techniques with neuromorphic architectures, enabling robots to adapt and learn from their experiences while exhibiting robustness and energy efficiency. The envisioned applications of our work span a wide range of domains, including autonomous navigation, object recognition, scene understanding, and human-robot interaction. By utilizing neuromorphic principles, we aim to empower robots with the ability to perceive and interpret their environment more efficiently, leading to improved decision-making, adaptability, and overall performance. In this talk, we will present our ongoing research efforts in developing neuromorphic-inspired algorithms for robotic perception. We will discuss the design considerations, experimental methodologies, and results obtained thus far, highlighting the advantages and challenges associated with adopting a neuromorphic approach. Furthermore, we will explore the potential impact of our findings on the development of next-generation robotic systems and their role in shaping the future of embodied neuromorphic AI. Keywords: neuromorphic engineering, robotic perception, embodied cognition, machine learning, computer vision, parallel computing, adaptive systems, intelligent robotics, neuromorphic hardware, artificial intelligence.
Dr. Fakhreddine Zayer is currently a Post Doctoral fellow in Electrical and Electronic Engineering at the Center for Autonomous Robotic Systems (KUCARS), Khalifa University, UAE, since 2021. He received his B.S. degree in physics in 2011 and his M.S. degree in materials, nanostructures, devices, and microelectronic systems in 2013 from the Faculty of Sciences of Monastir, University of Monastir, Tunisia. He received his Ph.D. degree in Electronics and Computer Engineering in 2021 from the National Engineering School of Monastir (ENIM) of Monastir University, Tunisia. He worked as a research associate with the System on Chip Center(SOCC), at Khalifa University in 2019-2021. Prior to joining Khalifa University, he worked as a senior staff Engineer in the integration of Materials to the system (IMS)-Thales group, France 2014-2016. His research focus is on Cutting-Edge AI Chips for AI Computing and Engineering approaches and the advances of computer vision and range sensing for Robotics. Several scenarios and algorithms are investigated such as navigation, path planning, localization, and control, etc, for next-generation robotics. The related AI Chip Technologies for Robotics include Video/Image, Sound and Speech, NLP, Control, NN topology, DNN and ML Algorithms, Chip Performance Optimizations, Programmable and neuromorphic chip, System-on-chip Architecture, Development Tool-chain, High Bandwidth off-chip Memory, High-Speed Interface, Bionic Devices, On-Chip Memory, and 3D Staking.
Brief Description on the talk: Learning with sparse rewards is usually inefficient in Reinforcement Learning (RL). Hindsight Experience Replay (HER) has been shown an effective solution to handle the low sample efficiency that results from sparse rewards by goal relabeling. However, the HER still has an implicit virtual-positive sparse reward problem caused by invariant achieved goals, especially for robot manipulation tasks. To solve this problem, novel model-free continual RL algorithms, such as Relay-HER (RHER) have been proposed. The latter is based on the decomposition and rearrangement of the original long-horizon task into new sub-tasks with incremental complexity. Subsequently, a multi-task networks are designed to learn the sub tasks in ascending order of complexity. To solve the virtual-positive sparse reward problem, this talk will target Random-Mixed Exploration Strategies, in which the achieved goals of the sub-task with higher complexity are quickly changed under the guidance of the one with lower complexity.
Dr. Benjamin C.K. Tee is appointed President’s Assistant Professor in Materials Science and Engineering Department at the National University of Singapore (NUS). He obtained his PhD (EE) at Stanford University, and was a Singapore-Stanford Biodesign Global Innovation Postdoctoral Fellow in 2014. He has developed and patented several award-winning electronic skin sensor technologies. He is an MIT TR35 Innovator (Global) in 2015 and Singapore National Research Foundation (NRF) Fellow. His research group aims to develop technologies at the intersection of materials science, mechanics, electronics and biology, with a focus on sensitive electronic skins that has tremendous potential to advance global healthcare technologies in an increasingly Artificial Intelligence (AI) and Robotics driven era.
Neuromorphic computing bears huge potential for many applications in robotics, with some of the main advantages being low-latency processing, support for online learning, and extremely low energy consumption. As more and more neuromorphic computing platforms are becoming available, the main challenge today lies in interfacing them with robotic systems and, most importantly, the development of new algorithms. This opens up a vast design space that is impossible to explore with traditional experiments on physical robots. In this talk, we highlight how virtual experiments performed in simulation enable the efficient design of neuromorphic perception and control algorithms for robotics. We provide an overview of the Neurorobotics Platform (NRP), our tool set for the modeling and simulation of brain-derived systems and neuromorphic algorithms. It is highly scalable and can interface with state-of-the-art neuromorphic processors. We further demonstrate how the NRP can be used to implement models and algorithms from computational neuroscience in robotics and present results from our research on neuromorphic computing in reinforcement learning and automotive sensor data processing.
Florian Walter received his Master’s degree in informatics with high distinction from Technische Universität München. During his Master studies, he completed an internship in the automotive industry and was a visiting student researcher in the Artificial Intelligence Laboratory at Stanford University where he worked in the field of online trajectory generation for robotics. In August 2014, Florian joined the HBP SP10 Neurorobotics research group as research assistant where he is now working in the field of neurobiological learning methods for robotics.
Memory architecture and design have been critical for digital systems to achieve ample storage, low latency, fast access time, and energy efficiency, especially for battery-operated devices. The increase of data generated by many devices such as mobile, sensors, communications, and security not only increased the requirements on memory capacity but also increased the challenges on memory access and energy. The memory interface has limited throughput and high latency, which has not been scaling at the same rate as data size or processing speed; this limits the performance of accessing the data, which refer to as the memory wall. In addition to the negative impact on latency and performance, large data movement results in high energy consumption. Research has been focusing on elevating the memory wall issue by engineering more memory hierarchy and increasing local on-chip memory. This has partially reduced the timing issue but did not address the high leakage and active energy consumption. It is estimated that more than 60% of energy spent on most computing platforms is spent on data movements and memory access. The new era of big data and artificial intelligence-based applications has increased the urgency to solve memory capacity, data movement energy, and memory wall issues. Some solutions have brought processing into centralized cloud computing, with high performance and large memory hardware capacity available. However, this brought a new challenge to communications, privacy, security, and latency, especially for real-time applications. The goal of this lecture is to highlight the after mentioned challenges and to present a new paradigm of computing beyond von Neuman's architecture to enable processing as close to the data source as possible. This includes in-memory computing, near memory computing architecture. Both existing and emerging memory technologies will be explored. Since the new computing paradigm is more data-centric than traditional processing-centric, the traditional single architecture for all applications is not feasible, but rather a domain-specific architecture and hardware solutions need to be adopted. Popular high computing functions such as Query, MAC, hamming distance, and image compression will be presented as an example of in-memory hardware accelerators.
Dr. Baker Mohammad is the director of the System on Chip center and professor of EECS at Khalifa University. Before joining Khalifa University, he was a Senior Staff Engineer/Manager at Qualcomm, Austin, Tx, USA, for 6-years, where he was engaged in designing high-performance and low-power DSP processors used for communication and multi-media application. Before joining Qualcomm, he worked for ten years at Intel Corporation on a wide range of microprocessors design from high-performance server chips > 100Watt (IA-64) to mobile embedded processors low power sub 1 watt (xscale). He has over 16 years of industrial experience in microprocessor design, emphasizing memory, low power circuit, and physical design. Baker earned his PhD from the University of Texas at Austin in 2008, his M.S. degree from Arizona State University, Tempe, and his BS degree from the University of New Mexico, Albuquerque, all in ECE. His research interests include VLSI, power-efficient computing, embedded memory and in-memory computing, neuromorphic computing, emerging technology such as Memristor, STTRAM, hardware accelerators for Cyber-Physical Systems and AI. He is also engaged in a microwatt range computing platform for wearable electronics and WSN, focusing on energy harvesting, power management, and power conversion, including efficient dc/dc, ac/dc converters. Baker authored/co-authored over 200 referred journals and conference proceedings, >5 books, >20 US patents, multiple invited seminars/panellists, and the presenter of >3 conference tutorials, including one tutorial on Energy Harvesting and Power Management for WSN at the 2015 (ISCAS). Baker is on the advisory board for the secure systems research center part of the Technology Innovation Institute. Baker is an associate editor for IEEE Transaction on VLSI (TVLSI), IEEE Access, and Scientific Reports journals. Dr Mohammad participates in technical committees at IEEE conferences and reviews for TVLSI, IEEE Circuits and Systems journals. He has received several awards, including the KUSTAR staff excellence award in intellectual property creation, IEEE TVLSI best paper award, 2016 IEEE MWSCAS Myrill B. Reed best paper award, and Qualcomm Qstar award for excellence in performance and leadership. SRC Techon's best session papers for 2016 and 2017 on the community.
Hyperdimensional computing (HDC) architectures in machine learning have gained unprecedented popularity in recent years due to their simplicity, robustness, and computational efficiency. Moreover, when compared with the deep neural networks, that are both computationally expensive and require exhaustive training rounds, HDC emerges as an excellent alternative that offers a highly competitive performance with extremely lower computational complexity as compared to the neural networks. Due to the optimal trade-off between computational efficiency and recognition performance, HDC showcases a strong potential to be deployed on a resource-constrained devices that are used in wide variety of applications related to robot sensing, perception, and manipulation. In this talk, we will discuss some of the advances in brain inspired HDC architectures to solve robotic manipulation tasks. More specifically, we will discuss some frameworks that improve the performance of soft-actor critic (SAC)-based reinforcement learning algorithms for grasping moving objects using HDC architectures, that are trained using trajectory-based contrastive unsupervised learning. Through simulation and real-world experiments, we will see how HDC models and trajectory-based learning can overcome the inherent limitations of sample inefficiency for the SAC-based RL agents towards grasping moving objects.
Dr. Taimur Hassan received the Ph.D. degree in computer engineering from the National University of Sciences and Technology, Islamabad, Pakistan, in 2019., He is currently Assistant Professor at Abu Dhabi University Center for Autonomous Robotic Systems (KUCARS) and the Center for Cyber-Physical Systems (C2PS), Department of Electrical Engineering and Computer Science, Khalifa University of Science and Technology, Abu Dhabi, United Arab Emirates. He has led many local and foreign research projects. His research interests lie in the fields of medical imaging, computer vision, deep learning, the Internet of Things, and robotics.,Dr. Hassan’s Ph.D. research won the Gold Award in the Research and Development Category at the Pakistan Software Houses Association for IT and ITeS (P@SHA) ICT Awards in 2016, the Gold Award in the Research and Development Category at the Asia Pacific ICT Alliance (APICTA) Awards in 2016, and the Gold Award in the Artificial Intelligence category at P@SHA ICT Awards in 2018. He was a recipient of many national and international awards.
Muaz Al Radi, Fakhreddine Zayer, Mahmoud Said Elmezain, Vidya Sudevan, Jorge Dias, Naoufel Werghi
Department of Electrical Engineering and Computer Science, Khalifa University, Abu Dhabi, UAE
Abstract: Due to their attractive features and operational flexibility, Unmanned Aerial Vehicles (UAVs) have received a lot of attention in recent years in a variety of engineering and practical applications. For the UAV system to function properly and effectively, suitable control systems are needed. One of the challenges in the implementation of autonomous UAV systems is the design of the onboard controller that controls the UAV’s movement in the environment and ensures navigation from the starting position to the goal position efficiently while ensuring no collisions with any obstacles in the environment. However, a challenge in the implementation of this approach is the limited computational power available onboard the UAV system, thus significantly limiting the size and complexity of models and algorithms that are applicable in real time. The use of computationally-efficient neuromorphic artificial intelligence models is a promising solution for onboard visual and multimodal data processing. In this work, an efficient intelligent control algorithm based on end-to-end Spiking Artificial Neural Networks for vision-based control of autonomous UAVs for autonomous navigation and obstacle avoidance is proposed. A Spiking Convolutional Neural Network (SCNN) was trained and evaluated for generating control commands based on the UAV’s video data stream. The method is tested and compared to conventional approaches in terms of prediction accuracy and computational efficiency.
Keywords: Unmanned Aerial Vehicle (UAV), Navigation and obstacle avoidance, Visual servoing, Simultaneous localization and mapping (SLAM), Spiking Neural Network.
Mariam Alzaabi, Fakhreddine Zayer, Naoufel Werghi, and Jorge Dias
Department of Electrical Engineering and Computer Science, Khalifa University, Abu Dhabi, UAE
Abstract—This paper presents an energy-efficient approach to dehazing underwater images using spiking generative adversarial networks (GANs). The hazy quality of underwater images impairs the visual performance of vision-based navigation systems in the underwater environment. Our solution employs deep neural networks inspired by the brain’s biological structures to restore clear underwater scenes from hazy images, leveraging the energy-efficient principles of neuromorphic computing. The results of our experiments demonstrate that this neuromorphic GAN-based model outperforms traditional learning-based dehazing methods, leading to clearer images and improved navigation accuracy. This research makes a valuable contribution to the development of energy-efficient neuromorphic solutions for underwater image dehazing, with significant implications for autonomous underwater navigation.
Keywords—Neuromorphic computing, image dehazing, underwater imaging, generative adversarial network.
Vidya Sudevan1 , Fakhreddine Zayer1 ,Mahmoud Elmezain1 , Mariam Alzaabi1 , Muaz Alradi 1 , Giulia De Masi1,2 , Jorge Dias1
1-Center for Autonomous Robotic Systems, Khalifa University, Abu Dhabi, UAE
2-Technology Innovation Institute, Abu Dhabi, UAE
Abstract—Unmanned underwater vehicles (UUVs) must be precisely positioned for oceanographic research and critical marine infrastructure inspections. Sensor fusion, which combines complementary data from multiple sensors, addresses the issue of localization in mobile robot applications. Because of their inherent power efficiency and impressive inference accuracy across various cognitive tasks such as image classification and speech recognition, Spiking Neural Networks (SNNs) are quickly emerging as promising candidates for brain-inspired neuromorphic computing. Recent SNN efforts have concentrated on implementing deeper networks with multiple hidden layers to incorporate exponentially more difficult functional representations. This paper proposes an end-to-end hybrid spike-based CNN-LSTM framework for predicting the 6D pose of underwater vehicles in an unstructured environment. The system is divided into three phases: (a) a spiking CNN network for image feature extraction, (b) a spike LSTM network for temporal feature extraction from the Inertial Measurement Unit (IMU), and (c) a direct fusion architecture for fusing and estimating the underwater vehicle’s position and orientation.
Index Terms—Spike-based CNN-LSTM network, hybrid framework, end-to-end model, 6D pose estimation, underwater vehicles
Mahmoud ElMezain1, Fakhreddine Zayer1, Vidya Sudevan1, Muaz Alradi1, Sajid Javed1, Federico Renda1, Giulia De Masi1,2, Jorge Dias1
1-Center for Autonomous Robotic Systems, Khalifa University, Abu Dhabi, UAE
2-Technology Innovation Institute, Abu Dhabi, UAE
Abstract—Tanks and underwater pipelines are utilized for the storage and long-distance transportation of fluids to a market location for consumption. If such structures leak, major issues including environmental contamination, explosion, financial loss, and so on may ensue. Therefore, it is crucial to use intelligent inspection technologies to guarantee the integrity of underwater structures. Remotely Operated underwater Vehicles (ROVs) have been commonly deployed by oil and gas companies for manual visual inspection of structures to detect cracks. In this study, we propose a deep learning vision-based method for the autonomous detection of cracks in underwater structures. Our deep learning approach is based on spiking Convolutional Neural Networks (CNN) using leaky-integrate-fire (LIF) neuron models. The images used for training have been pre-processed using a pre-trained Generative Adversarial Network (GAN) to add underwater noise, allowing the detection network to learn and extract features under settings of uneven light, high noise, and blurred images which are then encoded as spikes. Experiments are conducted by deploying an ROV in our facility’s pool in which an underwater structure with multiple cracks is placed. The robot scans the structure in a uniform manner, while the detection spiking CNN detects the damage to the structure and proceeds to display a bounding box surrounding the cracks.
Index Terms—Spike-based CNN Network, Crack Detection, Underwater Structures, Underwater Vehicles.
Fakhreddine Zayer, Rizwana Kausar, Jaime Viegas, Jorge Dias
Center for Autonomous Robotic Systems, Khalifa University, Abu Dhabi, UAE
Abstract: In recent years, the development of artificial intelligence (AI) and robotics has gained significant momentum. One of the challenges in this field is developing autonomous robots that can perform complex tasks in dynamic and unstructured environments. Olfaction sensing is a key capability required for robots to navigate, detect, and identify different types of objects and substances. However, current olfactory sensors are limited in their sensitivity and selectivity, making them inadequate for many real-world applications. In this paper, we propose a novel approach to olfaction sensing based on neuromorphic photonics and spike-based Bayesian learning algorithms. The proposed system mimics the functioning of the olfactory system in animals and is capable of detecting and identifying a wide range of volatile organic compounds (VOCs). The system is based on a hardware-friendly architecture that is highly efficient and can operate in real-time. The proposed algorithm uses Bayesian inference to estimate the probability of a particular odor based on the spiking activity of the photonic neurons. The algorithm is implemented on a custom-designed neuromorphic photonics chip, which enables fast and efficient processing of the sensory input. The system can also learn and adapt to new odors over time, making it highly versatile and adaptable. We demonstrate the effectiveness of the proposed system through experimental results on a robot platform. The robot is capable of detecting and identifying different types of VOCs in real-time, enabling it to navigate and interact with its environment autonomously. Overall, our approach offers a promising solution for developing highly sensitive and selective olfactory sensors for autonomous robotics applications. The use of neuromorphic photonics and spike-based Bayesian learning algorithms provides a hardware-friendly and low power solution that can operate in real-time and adapt to new environments and odors.
Keywords: Neuromorphic Photonics, Olfaction Sensing, Autonomous Robotics, A Spike-Based Bayesian Learning, Hardware-Friendly Architecture
Baker Mohammad received the B.S. degree from the University of New Mexico, Albuquerque, the M.S. degree from Arizona State University, Tempe, and the Ph.D. degree from The University of Texas at Austin, in 2008, all in ECE.
He is currently the Director of the System on Chip Center and an Associate Professor in EECS at Khalifa University. Prior to joining Khalifa University, he was a Senior Staff Engineer/the Manager at Qualcomm, Austin, USA, for six years, where he was engaged in designing high-performance and low-power DSP processors used for communication and multimedia applications. Before joining Qualcomm, he worked for ten years at Intel Corporation on a wide range of microprocessors design from high-performance, server chips more than 100Watt (IA-64), to mobile embedded processor low power sub one watt (xscale). He has over 16 years of industrial experience in microprocessor design, emphasizing memory, low power circuits, and physical design. His research interests include VLSI, power-efficient computing, high yield embedded memory, emerging technologies, such as memristor, STTRAM, in-memory-computing, and hardware accelerators for cyber-physical systems. He is also engaged in a microwatt-range computing platform for wearable electronics and WSN, focusing on energy harvesting, power management, and power conversion, including efficient DC/DC and AC/DC converters. He authored/coauthored over 150 refereed journals and conference proceedings, more than three books, more than 18 U.S. patents, multiple invited seminars/panelists, and the presenter of more than three conference tutorials, including one tutorial on energy harvesting and power management for WSN at the 2015 (ISCAS). He is an Associate Editor of IEEE Access and IEEE Transaction on Very Large Scale Integration (VLSI) Systems.
UAE, P.O. Box: 127788
Abu Dhabi
Phone: +971 2 312 4499
Email: baker.mohammad@ku.ac.ae
Dr. Benjamin C.K. Tee is appointed President’s Assistant Professor in Materials Science and Engineering Department at the National University of Singapore (NUS). He obtained his PhD (EE) at Stanford University, and was a Singapore-Stanford Biodesign Global Innovation Postdoctoral Fellow in 2014. He has developed and patented several award-winning electronic skin sensor technologies.
He is an MIT TR35 Innovator (Global) in 2015 and Singapore National Research Foundation (NRF) Fellow. His research group aims to develop technologies at the intersection of materials science, mechanics, electronics and biology, with a focus on sensitive electronic skins that has tremendous potential to advance global healthcare technologies in an increasingly Artificial Intelligence (AI) and Robotics driven era. He can be reached at www.benjamintee.com.
xxxxxxx
xxx
Phone: +xxx xxx xxx
Email: benjamin.tee@nus.edu.sg
Alois C. Knoll is a computer scientist and professor at Technical University Munich TUM. He teaches and conducts research in the fields of autonomous and embedded systems, robotics and artificial intelligence.
In 2011, he founded the interdisciplinary course "Robotics, Cognition, and Intelligence" at TUM, which has become one of the largest MSc programs in TUM’s CS Dept. In 2007 he became a member of the EU's highest advisory board for information technology, ISTAG. In this function, he was in involved in the development of the EU flagship projects and was ab author of the report "European Challenges and Flagships 2020". In 2009, he co-founded "fortiss", the Munich Institute for Software, which has since been transformed into a state institute of Bavaria. He has coordinated the project ECHORD++, an initiative to bring together the robotics industry and universities to speed up robot technology’s route to market. He has been head of the sub-project "Neurorobotics" of the EU flagship project “Human Brain Project”. Since 2011 he has been PI at TUMCREATE, a joint venture of NTU and TUM-Asia in Singapore. His focus there is on modeling, simulation and optimization for infrastructure, both in methodical development and in the construction of practical systems.
Technical University of Munich.
Germany
Phone:+49 (89) 289 – 18104
Email:knoll@mytum.de
Lorem ipsum dolor sit amet, consectetur adipisicing elit. Harum, et iste sint.
read moreLorem ipsum dolor sit amet, consectetur adipisicing elit. Harum, et iste sint.
read moreLorem ipsum dolor sit amet, consectetur adipisicing elit. Harum, et iste sint.
read moreTo favor the uptake and the building of a larger community of users and stakeholders of embodied neuromorphic intelligence, the neuromorphic community should focus on the design of modular and reusable sensing and computing modules. The standardization of a common communication protocol, has already enabled sharing of modules and systems. Open-discussions and open-source implementations of algorithms and dataset-sharing will promote the growth of the field. A milestone on this path will be the definition of a suite of benchmarks that can be used to quantitatively compare the features and benefits of different neuromorphic systems.
Neuromorphic circuits need to convert sensory signals into address-events for further processing. The computational neuroscience community has a unique opportunity to inspire and educate neuromorphic engineers by pointing out the principles and strategies that the nervous system uses to convert analogue inputs to spikes and encode sensory signals. Tight collaboration with the neuroscience community will lead to important improvements in neuromorphic sensing circuits. Similarly, this community can provide useful insights for designing recurrent Spiking Neural Networks (SNNs) composed of noisy and inhomogeneous circuits to carry out signal processing and computation. In this respect, it will be important to link specific neuroscience observations to their most basic computational role in order to isolate the basic mechanisms that are sufficient to implement a given functionality. The hardware implementation will then reproduce such a reduced “minimalist” model, where features, complexity, detail, and diversity have corresponding computational functions.
Emerging memory technologies, like memristors hold great promises for improving conventional computing architectures, However, they also represent an important opportunity for designing new types of solid-state nano-scale devices that could directly emulate the physics of real synapses, and therefore provide the computing substrate for implementing the principles of neural computation more efficiently. The material science community should therefore attempt to embrace and exploit the non-linear physics of these devices to optimize the design of embodied neuromorphic computing architectures.
Similar to how computers use a hierarchy of levels of abstraction to manage the definition of complex operations, computer science can leverage on the notions and tools developed so far to define new methods for combining neural computational primitives, to achieve intelligent functionalities. A challenge that lays ahead is also how to formalize computation using non-linear dynamics, stochastic, and probabilistic methods, including embodiment in the robotic platform.
As the neuromorphic approach is a good fit for complex systems where the control is non-trivial, it is a perfect match to soft robotics. There is a need for undefined, providing use cases to the neuromorphic community. The resulting perceptive and cognitive functions–implemented using neuromorphic computational substrates–must be embedded on robots, where the morphology of the platform can influence the way sensory signals are acquired (e.g. through a different placement of the sensors) and the way actions are executed (e.g. different kinds of locomotion, rigid versus soft actuation, etc.). Neuromorphic engineering, thanks to its ability to implement adaptive circuits and systems for solving non-linear control systems, can offer a solution to the complex control of soft robots.
Major events calendar 2023 - London Convention Bureau.
Lorem ipsum dolor sit amet, consectetur adipisicing elits. Fugit vel, eius cum eum.
Lorem ipsum dolor sit amet, consectetur adipisicing elits. Fugit vel, eius cum eum.
Lorem ipsum dolor sit amet, consectetur adipisicing elits. Fugit vel, eius cum eum.
Khalifa University Center for Robotics and Autonomous Systems (KUCARS), Khalifa University.
Professor, Center for Robotics and Autonomous Systems (KUCARS), Khalifa University.
Assistant Professor, Head of the Artificial Intelligence and Data Analysis Lab- DIEI .
Professor, Center for Robotics and Autonomous Systems (KUCARS), Khalifa University.
121 King Street, Melbourne Victoria 3000 Australia