Name: Dr Chapa Hewa Pelendage
Academic Division: Civil Engineering
Research Group: Bio-Inspired Robotics Lab
Fellowship period: 15 June 2023 – 14 June 2026
Email: csh66@cam.ac.uk
https://orcid.org/0000-0003-3055-4920
Research Project: A Multi-Agent System for Heavy Machine Operation through Context-Aware Sensor Fusion
Abstract:
Construction sites in the UK remain hazardous, responsible for over 20% of workplace fatalities. These incidents result not only in tragic loss of life but also in significant financial costs due to equipment damage and compensation claims. A key factor in many accidents is the limited situational awareness of heavy machinery operators and workers. There is a critical need to integrate Robotics, and Automation to improve safety on these sites, especially where automation is challenging due to the complexity and variability of the environment.
This project follows a three-stage approach to develop a smart, adaptable safety system for construction.
First, the project will identify specific safety requirements for construction workers by analysing common site hazards. Based on this, a digital interface will be developed, enabling workers to monitor risks in real-time. Next, an intelligent algorithm will allow the system to interpret construction site activity as work proceeds, enhancing its ability to anticipate risks and respond to dynamic changes. Finally, visual and audio alerts will be incorporated to deliver immediate feedback on hazards, making the system practical for high-traffic, limited-visibility areas.
Field testing will help refine this technology to address the unique safety needs of construction environments. Unlike manufacturing, where robotic systems are well-integrated, this project represents one of the first targeted applications of robotics for safety on construction sites. Once successful, this solution can transform industry practices, contributing to sustainable, resilient infrastructure and aligning with SDGs focused on safety, innovation, and sustainable cities.
Upon completion, this research will be deployed as a human-machine interface in work environments, control rooms, and on heavy machinery as a real-time tool to observe otherwise hidden areas, anticipate hazards, and help avoid them—all without replacing human oversight. The system’s modular design will enable it to be adapted to different work sites.
This research establishes a smart interface between workers and machines, combining robotics and algorithms to manage complex, dynamic environments. For instance, if a machine operator needs an overview of a distant area, they can request a robot to capture a real-time image, ensuring safer conditions. Additionally, algorithms will support machine-to-machine communication, creating a network of "talking machines" that share critical information. Concepts like reinforcement learning, self-organisation, and embodied intelligence will make this a robust, responsive safety system.
Current Progress: Development has reached the simulation stage, involving heavy machinery, sensorized environments, multi-agent system (MAS) algorithms, and human factors such as emotions that could serve as cues to prevent hazards. Currently, the fellow is seeking an industry secondment to observe worker interactions with these systems, evaluate the feasibility of using robots in dynamic settings like construction and test the simulations in real world. Addressing these human-centred requirements will strengthen the system’s integration and adaptability, ensuring a safe, collaborative, and practical solution for real-world applications.