Infrastructure Robotics

Automatic Robot System for Indoor High Rise Spray Painting

High ceiling painting is inevitable & common in building construction, especially in industrial workshop. Its painting takes a lot of effort for details and is no doubt time consuming. Traditional high ceiling painting is manually done by means of ladders or hydraulic lifts and often results in unreliable painting quality. Moreover, the high-place operation (up to 10 meters) will create lethal danger to painting workers. This project aims to develop a mobile robot system equipped with a novel long reach mechanism for high ceiling and wall painting applications.

Co-development of Light-Industrial Use Multiple-axis Robotic Spray-painting System a.k.a. Outdoor Wall Cleaning & Painting Robotics System

Currently, outdoor wall painting process on buildings is done manually by painting workers who are carried by a manually controlled gondola platform. For each horizontal axis movement, worker will need to shift the gondola manually and the whole painting process becomes time-consuming. This is also hazardous as the painter may fall off from a great height. This project aims to develop an automated outdoor wall cleaning and painting robotics system, named Spiderman. When successful, it will provide cleaning and painting services on outdoor walls which such as HDB flats and high-rise buildings.

Principal Investigator: Professor Chen I-Ming

• AI-driven: iScan2BIM is driven by our Machine Learning, Deep Learning, and Bayesian Network algorithms for direct and automatic conversion of building information models from laser-scanned point cloud.
• Robotics-powered: Generation of an auto-fused and single point cloud dataset with the aid of our BIMbot (Giraffe) through automatic and optimal LiDAR scanning of multi-modal images.
• VR-enhanced: VR function for immersive visualization and interaction of the BIM reconstructed from point cloud onsite after scanning.
• Cloud-integrated: Can be run anytime, anywhere by anyone using a mobile device at frontend supported by an AWS platform at backend.

Principal Investigator: Associate Professor Cai Yiyu

Deploying AI foundation models on resource-constrained edge devices is crucial for real-world robotic applications. Our research focuses on designing algorithms to minimize data and computational costs while preserving model accuracy. We develop efficient model compression, quantization, and optimization techniques to enable foundation models to run efficiently on edge hardware for robotic perception, planning, and control. Our ultimate goal is to create on-chip foundation models tailored for specific robotic tasks, ensuring scalable, low-latency, and energy-efficient AI inference.

Principal Investigator: Assistant Professor Yang Jianfei