PhD Studentship: Neuromorphic Edge AI for Energy-Efficient Robotics

About the Project

This PhD project is built on a multidisciplinary collaborative project between Anglia Ruskin University (AI) and University of Cambridge (Electronic Engineering Division in the Department of Engineering). This project will investigate the convergence of Edge AI and Neuromorphic Computing to enable adaptive, low-power robotic systems.

Supervision Team:

• Professor Yonghong Peng (Anglia Ruskin University, Professor of Artificial Intelligence, https://www.aru.ac.uk/people/yonghong-peng)
• Professor Andrew Flewitt (University of Cambridge, Professor of Electronic Engineering, https://www.eng.cam.ac.uk/profiles/ajf23)

Background and Motivation:

As robots become increasingly integral in society, they must adapt to changing environments and cooperate with human partners effectively. Traditional AI systems, such as neural networks and deep learning algorithms, have excelled in tasks like computer vision, image recognition and large language models (LLM). However, their reliance on extensive computational resources results in excessively high energy consumption, making them unsuitable for energy-constrained applications such as edge devices and social robots. If the energy issue is not addressed, it will become a major constraining factor for the continued advancement of AI.

Neuromorphic computing offers a promising approach for innovation in this space. Neuromorphic systems, processing information using spiking neurons and synapses, enable energy-efficient, brain-like decision-making capabilities. This project will aim to develop effective approach to enable the integration of neuromorphic electronics/computing and edge AI to support low-power robotic systems that sense, think, and act efficiently in real-time and real-world environment. In particular, the project will investigate the opportunity for embedding AI processing around a robotic system. For example, a gripper coated with a flexible ‘smart skin’ could have local processing to interpret data so reducing the data that needs to be sent for central processing, thereby making the system more efficient both in terms of hardware and energy.

Research Aims and Objectives:

The project will aim to create energy-efficient neuromorphic systems to enable AI on edges through the following objectives:

• Advancing Neuromorphic Electronics/Computing: Develop advanced neuromorphic architectures that simulate neural and synaptic structures of the brain to improve adaptability and real-time decision-making in robots. There will be a particular focus on the use of flexible electronics to achieve this based on thin film transistor (TFT) logic.
• Innovating Edge AI: Implement Edge AI algorithms optimised for neuromorphic systems, allowing robots to process data locally, with high security, in resource-constrained environments, enhancing their autonomy and efficiency, tailored to the hardware solutions developed in Aim 1.
• Implementing Energy-Efficiency in AI and Robotics: Leverage the low-power characteristics of neuromorphic architectures to design adaptive AI systems for energy-efficient, real-time operation on edge devices. Investigate how neuromorphic perception can be integrated with motor control using flexible electronics to enable robots to interact intelligently with their environment while using minimal energy.

Methodology and Research Plan:

• Literature Review and Critical Analysis: Review and analyse existing research on neuromorphic computing and flexible electronics and identify gaps to propose novel solutions. (months 1-6)
• Neuromorphic Computing Development: Collaborate with Professor Andrew Flewitt research lab to design neuromorphic circuits using flexible TFT-based logic incorporating spiking neurons and synaptic architectures using flexible electronic materials. Develop use cases tailored to flexible robotics, optimising energy efficiency and adaptability. (months 3-18)
• Edge AI Advancement: Design and implement AI algorithms for neuromorphic systems, focusing on real-time training and operation to enhance adaptability and safety. Advance neuromorphic-enabled machine learning for multimodal sensory data to enable robots to recognise patterns, learn behaviours, and interact with their environment. (months 6-27)
• Robotics Innovation: Develop use cases for a robotic platform integrating neuromorphic computing and machine learning. Test the platform in scenarios requiring human-AI cooperation, such as dynamic human-robot interactions, and evaluate the system's energy consumption, adaptability, and human cooperation performance. (months 15-36)

Expected Outcomes:

This project will contribute to advancements in both neuromorphic computing/engineering and AI for robotics, opening new avenues for developing energy-efficient, adaptive systems that can seamlessly integrate into real-world applications. Research outcomes include:

• A comprehensive literature review to inform the development of a novel framework that combine AI and neuromorphic computing at edges for flexible robotics. (Year 1)
• A highly energy-efficient robotic testbed that is capable of real-time adaptation for real-world applications with significantly lower power consumption compared to traditional AI systems. (Year 2)
• Testing and evaluation of the edge AI capabilities of the robots in dynamic environment such as human-robot collaboration. (Year 3)
• Publications in leading AI, robotics, and neuromorphic computing conferences and journals. (Year 1/2/3).

You will demonstrate excellent knowledge and skills in

• Electronic Engineering or Computing Engineering
• Advanced mathematics knowledge
• Cutting edge programming skills
• Analytics and problem-solving skills

Qualifications:

Applicants should have a minimum of a 2.1 Honours degree in a relevant discipline. An IELTS (Academic) score of 6.5 minimum (or equivalent) is essential for candidates for whom English is not their first language.

In addition to satisfying basic entry criteria, the University will look closely at the qualities, skills, and background of each candidate and what they can bring to their chosen research project in order to ensure successful and timely completion.

Any additional qualifications

You will demonstrate excellent knowledge and skills in

• Engineering
• Mathematics
• Cutting edge Programming
• Analytics and Problem Solving

How to apply:

To apply, please complete the application form available from the following website: Computing and Information Science - MPhil, PhD - ARU via the above ‘Apply’ button. Please ensure the reference ‘PhD Studentship: Neuromorphic Edge AI for Robotics’ is clearly stated on the application form, under the title ‘Outline of your proposed research’. Within this section of the application form, applicants should include a 500-word outline of the skills that they would bring to this research project and detail any previous relevant experience. 

Interested applicants should direct initial queries about the project to Professor Yonghong Peng via email: Yonghong.Peng@aru.ac.uk, or Professor Andrew Flewitt (University of Cambridge). For enquiries regarding the process and eligibility please contact SE-Research@aru.ac.uk.

Interviews are scheduled to take place in May 2025

We value diversity at Anglia Ruskin University and welcome applications from all sections of the community.

Closing Date: 09 May 2025

Funding Notes

A 3-year studentship is offered, intended to start in Sept 2025, providing a tax-free stipend of £19,237 per annum plus tuition fees at the UK rate. Due to funding restrictions, this studentship is only available as a full-time position and to UK candidates. 

Project location: Cambridge campus.

Candidates for this PhD Studentship must demonstrate outstanding qualities and be motivated to complete a PhD within 3 years.  

References

Bartolozzi, C., Indiveri, G. & Donati, E. Embodied neuromorphic intelligence. Nat Commun 13, 1024 (2022). https://doi.org/10.1038/s41467-022-28487-2
Krauhausen, I., Griggs, S., McCulloch, I. et al. Bio-inspired multimodal learning with organic neuromorphic electronics for behavioral conditioning in robotics. Nat Commun 15, 4765 (2024). https://doi.org/10.1038/s41467-024-48881-2
Yao, M., Richter, O., Zhao, G. et al. Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip. Nat Commun 15, 4464 (2024). https://doi.org/10.1038/s41467-024-47811-6
Yang, Y., Bartolozzi, C., Zhang, H., Nawrocki, R.A., Neuromorphic electronics for robotic perception, navigation and control: A survey, Engineering Applications of Artificial Intelligence, 126, Part A, 106838 (2023). https://doi.org/10.1016/j.engappai.2023.106838.