Skip to main content

Cognitive Robotics & Autonomous Systems

Cognitive Robotics

The term "cognitive robotics" is used to refer to robots with higher level cognitive functions that involve knowledge representation and reasoning. Several projects are currently undertaken in this area in collaboration with psychologists and neuroscientists from European institutions such as IIT in Genoa and Ferrara, Italy.

Our research consist of endowing robots with some cognitive capabilities which are key elements to autonomous systems, such as perception processing, attention allocation, anticipation, planning, reasoning about other agents, and perhaps reasoning about their own mental states.

Examples of projects undertaken in the lab in this area are:

  • Project carried out as a part of ASTRAEA T7 to develop a mechanism to present test cases to the ASTREA vehicle model for the handling of fuel system failure events within an open architecture. This project was critical for the whole internal integration of the ASTRAEA T7.  
  • Irrational Swarm: this project consists of endowing each agent/robots within the swarm with some irrational behaviour in order to mimic the human decision. The developed technique uses some psychological principles to model how the perceptions of human beings influence their choices from the given options.  
  • Probabilistic fuzzy reinforcement learning in social collaborative learning. This undergoing project is to use a hybrid algorithm and self-organizing particle system, enabling several robotic agents to learn an optimum behaviour combining their acquired knowledge about the environment to accomplish a given task, while dealing with the dimensionality problem of a real world environment, and its inherent uncertainties.  
  • Language grounding using the humanoid robot iCub platform. The main focus of this project is the investigation of the relationship between language and action.  

Robot CubTwo RobotCubs


We have actively participated in the development of the state-of-the-art European cognitive platform "humanoid robot" RobotCUB. The "iCub" is a child-like crawling robot that resembles a two-and-a-half-year-old child. The ultimate goal of this project is provide the cognition research community with an open human-like platform for the understanding of cognitive systems through  the study of cognitive development. This work is supported by the European Commission FP6, Project IST-004370. Researchers at Salford developed the mechanics, electronics and control of the spine and legs of the iCub. The complete development leaded to numerous publications, in the mechanics as well  as in control systems topics.

Additionally, our resources include NAO humanoid robots (Aldebaran Robotics) for research and development of a wide range of algorithms, ranging from modelling Cognitive functions of cognitive robotics to walking algorithms and visual signal processing.


Gorilla RobotTraditional robot design has been concerned primarily with the development of structures and mechanisms that have high accuracy and speed but at the expense of high mass and  power requirements and limited human interaction. Recent advances in computational power however have allowed lightweight and highly flexible structures, similar to those found in biological creations, to be used in robot design. This has led to the development of bio-mimetics were the trend is to try  to emulate the 'soft' compliant structure of muscle, bone, tendons and skin and combine this with the power, robustness, accuracy, and endurance of mechanical drives. This work used pneumatic Muscle Actuators (pMAs) as a soft actuators that can macroscopically replicate much of the action of natural  muscle. The actuators were tested in antagonistic pairs and used to power a robot primate (dimensionally comparable with a female gorilla) with a mass of less than 25kg constructed using light flexible materials.

Humanoid Robotic Hand

The University of Salford has a long history in the development of humanoid robotic hands. This particular hand was constructed for and installed at the Museum of Science and Industry in Manchester. The vast majority of dextrous robot hands are used in laboratory environments and so long term reliability is rarely a major design consideration. However, reliability is critically important for a museum exhibit as it  is not acceptable to have a display out of order for long periods of time. The hand designed in this work therefore had to be very reliable. The hand has a total of 15 degrees of freedom and was mounted to a 2 degree of freedom arm. An interface was produced allowing visitors to the museum to interact with the robot. This allowed them to use the robot to manipulate a ball and play simple tunes on a piano keyboard.

Robotic Hand

Swarm intelligence and multi-agent systems

Researchers in this area are concerned with the building of an Intelligent Collaborative Behaviour using Multi-Agent Systems/robots using  novel swarm intelligence techniques. We have introduced the irrationality theory for the first time to swarm optimisation techniques and to swarm robotics. This concept was applied in wide range of applications such as path planning, obstacle avoidance and emergent behaviours, using state of the art Khepera robots (K-Team) for testing our algorithms as well as the widely used simulation software Webots (Cyberbotics) for simulating and testing our designed robot models. This work has been supported by the Eu-Cognition Network 2008.

Projects include:

  • Study and implementation of particle-based simulation algorithms, which can be used for simulating viscoelastic behaviours for particles systems. These implementations could be reused in several types of particles behaviour simulations; thereby the solution itself could be tuned specifically for a certain type of system or very easily adapted for others;  
  • Crowd behaviour modelling;  
  • Space and security robotics: space repair and assembly, planet exploration, environmental surveillance and monitoring, costal and sea patrol, fire and bomb fighting, surveillance of critical infrastructure, search rescue missions;  
  • Swarm agents for CCTV Monitoring: Collaborate between CCTV's in order to make the monitoring effective (with Meyertech Ltd);  
  • Swarm cognitive agents for autonomous computing over the network;  
  • Activity Level Monitoring System;  
  • Capture illegal processes (games, chat, multi-media application etc) - Internet Monitoring;  
  • Application management;  
  • Install/Uninstall applications on remote machines autonomously and silently.

Swarm - 4 units and robot dogSwarm 2 - 5 units