Cognitive Robotics and Autonomous Systems (CoRAS) Lab

The Unitree G1 Edu Ultimate A humanoid robot is a compact, AI-driven humanoid platform designed for dynamic movement and precise manipulation, equipped with 43 joints for human-like mobility. It integrates learning-based control (imitation and reinforcement learning) with advanced perception through depth cameras and LiDAR for autonomous operation. The robot also features dexterous three-finger hands with force-controlled manipulation.

The AgileX Cobot S Kit is a ROS-based mobile manipulation robot that integrates an omnidirectional Ranger Mini 3.0 mobile base, a 6-axis Piper robotic arm with gripper, and a multi-sensor perception system into a single research platform. Designed for embodied AI and robotics experimentation, it supports autonomous navigation, real-time 3D mapping, and object manipulation in dynamic environments, powered by onboard computing for AI workloads.

Our Unitree Go2 “Stompy” robot dog is an agile, AI-powered quadruped robot designed for dynamic locomotion and intelligent interaction, weighing around 15 kg and capable of running at speeds of up to 5 m/s. It uses advanced 4D LiDAR and onboard AI to perceive its environment, enabling autonomous navigation, obstacle avoidance, and terrain adaptation, using reinforcement learning to perform complex movements, such as climbing, jumping and stable traversal.

Pepper is a pioneering social humanoid robot developed to facilitate human-robot interaction. Pepper integrates advanced technologies for people tracking and emotion recognition. It is a versatile platform for research and applications in social robotics.

Nao is a small humanoid robot used in research, education, and social interaction studies. It is equipped with sensors, cameras, and microphones to perceive its environment. Nao communicates through speech, gestures, and programmable actions. Its design makes it useful for studying robotics in learning and autonomous behaviour.

Amy A1 is an android-based service robot designed for interactive and assistive applications. It features advanced speech recognition, autonomous navigation, and multi-mode information display.

Buddy is an android-based social robot designed to foster connections with its expressive face and ability to display a range of emotions, creating natural and engaging interactions. It serves as a versatile assistant for education, entertainment, and human-robot interaction.

Q.bo One is an open-source robot built on Raspberry Pi and Arduino. It features stereoscopic vision, omnidirectional voice interaction, LED-based speech synchronisation, and touch sensors. It supports both beginner-friendly programming with Scratch and customisation for advanced projects using Python programming.

The TurtleBot3 Burger is a compact, affordable, and customisable ROS-based mobile robot. It features LiDAR-based technology supporting SLAM (Simultaneous Localisation and Mapping), autonomous navigation and person-following. Its modular design allows further expansion through additional sensors and computing hardware. Our fleet comprises four TurtleBot3 Burger robots, equipped with LiDAR sensors and augmented with depth cameras, such as the Microsoft Azure Kinect. The lab has three Azure Kinect depth cameras available, enabling enhanced environmental perception and depth-based sensing capabilities.

The TurtleBot3 Waffle (Raspberry Pi 4) is a variant of the TurtleBot3 mobile robot, equipped with enhanced features like an RGB-depth camera to support object detection/recognition and feature-based visual SLAM (VSLAM), alongside laser-based SLAM using its 360-degree sensor. Its increased payload capacity enables the integration of an OpenManipulator robotic arm and gripper, enhancing its completeness as a service robot with manipulation functions.

The laboratory is equipped with a motorised conveyor belt system designed to support research and experimentation in robotic manipulation, object detection, recognition and sorting. The conveyor features adjustable side guides to control object movement and maintain alignment along the belt. It can accommodate objects of different shapes, sizes and materials, enabling experiments in visual perception, object tracking, pick-and-place operations and automated sorting. The system can be integrated with robotic platforms (like G1, CobotS)and vision sensors to investigate perception–action coordination and autonomous object handling in dynamic environments.

The Microsoft HoloLens 2 is a standalone mixed-reality headset that combines holographic displays with spatial mapping, hand tracking, eye tracking and gesture recognition. It enables users to interact with digital content overlaid onto the physical environment, creating opportunities for human–robot interaction and collaborative robotics research. In robotic applications, it can be used to visualise robot states, trajectories, sensor data and virtual objects within the real workspace, as well as to provide intuitive, gesture-based interfaces for robot control. It also supports the development of mixed-reality scenarios in which humans and robots interact with shared virtual and physical objects, enabling research into human–robot collaboration, teleoperation, task planning and situated interaction.