An investigation into novel strategies for adaptive, intelligent and robust control of autonomous robots through the study of dynamic interactions bet
Lead Research Organisation:
University of Strathclyde
Department Name: Design Manufacture and Engineering Man
Abstract
The specific aim of my PhD project is to develop novel strategies for adaptive, intelligent and robust control of autonomous robots through the study of dynamic interactions between robots and their harsh environments. Through these advanced control strategies, the problem of efficiently dealing with the dynamic interactions between autonomous robots and their environments may be solved. The novelty of the control strategies lies in its use of information from on-board robotic vision systems and multiple sensory devices, equipped for the purpose of intelligent task planning and object manipulation. These advanced control strategies will be developed together with other sensor-based control techniques, such as hybrid force-position control.
In particular, this project aims to address the challenging issues associated with the underlying adaptive, intelligent and robust control strategies for dynamic interactions between autonomous robotic systems and the harsh environments in which they operate. The objectives of the project are therefore as follows:
1. Development of advanced control and communication prototyping platform for autonomous robotic applications that will enable the investigation of adaptive, intelligent and robust control strategies for autonomous robots in harsh environments;
2. Refinement of communication and sensing system to meet the stringent flexibility and reliability requirements needed for robust and adaptive robotic applications, including capabilities to monitor operational consistency and detect errors;
3. Development of adaptive, intelligent and robust control strategies for autonomous robots in hash environments, with particular focus on solving the problems of uncertainty and impact effects;
4. Implementation of the above control strategies on the developed control and communication platform to deliver a working system installed and running within a lab demonstration environment.
In particular, this project aims to address the challenging issues associated with the underlying adaptive, intelligent and robust control strategies for dynamic interactions between autonomous robotic systems and the harsh environments in which they operate. The objectives of the project are therefore as follows:
1. Development of advanced control and communication prototyping platform for autonomous robotic applications that will enable the investigation of adaptive, intelligent and robust control strategies for autonomous robots in harsh environments;
2. Refinement of communication and sensing system to meet the stringent flexibility and reliability requirements needed for robust and adaptive robotic applications, including capabilities to monitor operational consistency and detect errors;
3. Development of adaptive, intelligent and robust control strategies for autonomous robots in hash environments, with particular focus on solving the problems of uncertainty and impact effects;
4. Implementation of the above control strategies on the developed control and communication platform to deliver a working system installed and running within a lab demonstration environment.
Organisations
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/N509760/1 | 01/10/2016 | 30/09/2021 | |||
| 1803462 | Studentship | EP/N509760/1 | 01/10/2016 | 31/03/2020 | Cuebong Wong |
| Description | This research has investigated the problem of robotic planning for dynamic environments, where a robot is required to perform a set of tasks within environments that are either partially known, involve a degree of uncertainty, or are changing unpredictably over time. The planning problem involves finding a task plan that instructs a robot to perform a sequenced set of actions to accomplish the task. Additionally, we seek to obtain high quality solutions based on optimisation criteria such as minimising the execution time or maximising the distance away from hazards in the environment. This research has led to a number of developments that push the state-of-the-art in task planning and motion planning in robotics, where motion planning seeks to find a collision-free path to enable the robot to navigate safely between two points in its environment and task planning is concerned with finding a high level set of actions to accomplish a given goal. In this work, algorithms were firstly developed to address common planning problems encountered by mobile robots and robotic arms as a static problem. These algorithms enabled more efficient planning together with the capability to find high quality plans when compared to existing approaches. These algorithms were then extended to address the dynamic variant of these problems involving dynamic changes in problem variables. Until now, these dynamic planning problems have not been considered at the level of task planning within the large body of literature on robotic planning. For mobile robot planning, we consider the problem of finding an optimal task plan to visit a discrete set of points in the environment and to perform specific actions at these locations while avoiding collision with obstacles that occupy the same environment. The dynamic case additionally considers initially unknown or dynamic obstacles that are perceived during execution. For the planning of robotic arms, we specifically focus on the problem of robotic task sequencing, which involves finding an optimal sequence of motions to visit a large set of task points while avoiding collision. The problem is particularly challenging as a typical industrial robot containing 6 degrees of freedom can reach any single point in space in multiple ways, just as a human can reach a point in space with their arm in different postures. The complexity is further enhanced by the necessity to account for collision avoidance when optimising the sequence of motions. The dynamic case also extends the problem to consider dynamic changes to obstacles in the environment. The algorithms developed in this work uniquely solve the problems using an online planning approach, as opposed to the conventional method of finding solutions offline. This offers the potential to enable adaptive robotics by re-planning online to cope with changes that affect the optimality of the solution and the safety of the robot for a previously computed plan. |
| Exploitation Route | Adaptive robotics provides key capabilities to address many problems surrounding the deployment of robotics for industrial applications beyond traditional manufacturing. Challenging applications such as nuclear decommissioning, in-service inspection, and collaborative manufacturing between human and robots introduce high levels of unpredictability and uncertainty to the task that often prevent the use of autonomy due to the robot's inability to cope with changing task parameters. This work demonstrates the potential for overcoming these challenges through the use of planning techniques and indeed provides a number of methods for solving industrially relevant planning problems efficiently while generating useful, low-cost plans. The work conducted in this research additionally serves as a backbone for further developing robotic planning capabilities for optimal task planning and adaptive online re-planning. |
| Sectors | Aerospace, Defence and Marine,Agriculture, Food and Drink,Construction,Energy,Environment,Manufacturing, including Industrial Biotechology,Transport |
| Description | As discussed in the section for engagement activities, this research has led to a demonstration given at a National Nuclear Laboratory facility to key stakeholders within the nuclear sector. This demonstration serves an important purpose in showcasing the potential for adaptive robotics to change the way robots are used within nuclear decommissioning, which led to a change in mindsets for decision-makers. This is an important first step in changing the perspective and opinion of both professionals and the public on the role of robotics in society. Nuclear decommissioning remains a key priority within the UK, with key activities projected for the next 100 years. The opportunity to demonstrate the findings that have stemmed from this research is crucial to the acceleration of the development of robotics towards deployment. |
| First Year Of Impact | 2019 |
| Sector | Energy |
| Impact Types | Societal |
| Description | Advanced Forming Research Centre Route to Impact |
| Amount | £29,978 (GBP) |
| Organisation | University of Strathclyde |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 10/2017 |
| End | 04/2018 |
| Description | Demonstration at National Nuclear Laboratory |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | The activity consisted of a live demonstration of robotic capability developed as part of this research to stakeholders within the nuclear industry. These involve practicing professionals who make key decisions within various activities within the nuclear sector. The demonstration was presented to the audience at the National Nuclear Laboratory with the intention of highlighting the current possibilities of robotics for safer and more effective nuclear decommissioning. A publicity video was produced by the National Nuclear Laboratory that is freely accessible online to showcase the work. |
| Year(s) Of Engagement Activity | 2019 |
| URL | https://www.youtube.com/watch?v=eq80gH0zCv8&feature=emb_title |
| Description | University of Strathclyde Engage Week Knowledge Exchange Seminar - Flexible and Intelligent Industrial Robots for Applications in Autonomous Manufacturing |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Industry/Business |
| Results and Impact | The purpose of the seminar was to provide a knowledge exchange opportunity between industry and academia through a presentation on the outputs of the project on "Flexible and Intelligent Path Planning and Control of Industrial Robots towards Autonomous Hot Forging in the Digital Manufacturing Age", which was funded by the University of Strathclyde's Advanced Forming Research Centre (Route to Impact) and closely relates to the objectives of the award. Attendees from various industrial parties including KUKA, a major supplier of robotics, who took part in a panel discussion that followed the presentation. A live demonstration of the adaptive robot capabilities was also given to attendees at the AFRC. The seminar and demonstration collectively provided the audience with a new view of possible capabilities for robots in an industrial context. |
| Year(s) Of Engagement Activity | 2018 |