Towards Autonomous Robotic Systems Conference (TAROS)
September 8-10, 2021
TAROS 2021 will be virtual on Zoom and Discord. Register now to receive the links to participate for free in September closer to the event itself. We are looking forward to welcoming you to virtual TAROS 2021.
TAROS is the longest-running UK-hosted international conference on Robotics and Autonomous Systems (RAS), which is aimed at the presentation and discussion of the latest results and methods in autonomous robotics research and applications.
TAROS offers a friendly environment for robotics researchers and industry to take stock and plan future progress. It welcomes senior researchers and research students alike and specifically provides opportunities for research students and young research scientists to present their work to the scientific community.
As in previous years, the TAROS 2021 proceedings will be published in the Springer Lecture Notes in Artificial Intelligence series.
28 May 2021 – Submission deadline (5pm BST) 1 Jul 2021 – Review decision notification 28 Jul 2021 – Camera-ready paper submission (5pm BST)
- 8-9 Sep 2021 – Main Conference (see programme below)
- 10 Sep 2021 – Research and Industry Showcase Day
All times are British Summer Time (UTC+1).
Wednesday 8 September 2021
|10:00-10:30||Joining Online||Join us for TAROS 2021 on Zoom and Discord (bring your favourite beverage!)|
|10:30-10:45||Welcome talk||Simon Parsons & Charles Fox|
|10:45-11:00||Introduction to TAROS technology||Marc Hanheide|
|11:00-12:00||Keynote (Chair: Marc Hanheide)||Yiannis Demiris – Towards Personal Assistive Robots|
|12:00-12:20||ORAL1 (Chair: Heriberto Cuayahuitl)||Abdalkarim Mohtasib, et al. – A Study on Dense and Sparse (Visual) Rewards in Robot Policy Learning|
|12:20-12:40||Xintong Yang, et al. – An Open-Source Multi-Goal Reinforcement Learning Environment for Robotic Manipulation with Pybullet|
|12:40-13:00||Luigi Campanaro, et al. – CPG-Actor: Reinforcement Learning for Central Pattern Generators|
|13:00-14:00||lunch/poster||Enjoy the poster showreel and some food/drink|
|14:00-14:20||ORAL2 (Chair: Sergi Molina)||Karoline Heiwolt, et al. – Deep semantic segmentation of 3D plant point clouds|
|14:20-14:40||Muhammad Sami Siddiqui, et al. – Discovering stable robot grasps for unknown objects in presence of uncertainty using Bayesian models|
|14:40-15:00||R. Zhang, et al. – Improving SLAM in Pipe Networks by Leveraging Cylindrical Regularity|
|15:00-16:00||POSTERS1 (Chair: Marc Hanheide)||First poster session|
|16:00-16:20||ORAL3 (Chair: Taeyeong Choi)||James R. Hesselden and Gautham P. Das – CRH*: A Deadlock Free Framework for Scalable Prioritised Path Planning in Multi-Robot Systems|
|16:20-16:40||Gizem Ateş, et al. – Human-Robot Cooperative Lifting using IMUs and Human Gestures (swapped with next talk)|
|16:40-17:00||Elnaz Shafipour and Saber Fallah – Task-based ad-hoc teamwork with Adversary (swapped with previous talk)|
|18:00-19:00||IET Public Lecture (Chair: Simon Parsons)||Nick Hawes – An Uncertain Mission: Decision Making for Autonomous Robots.|
Towards Personal Assistive Robots
Yiannis Demiris, Imperial College London
How can robots assist people in achieving their tasks in a way that adapts to the individual? From physical assistance with activities of daily living to educational support, personalisation mechanisms share perceptual, representational, and learning challenges in determining the nature and timing of the optimal assistance by the robot: robots need to understand the current state and intention of their users through multimodal perception and action understanding algorithms, and adapt the assistance to the users’ particular (and evolving) needs. In this talk, I will present our research towards designing and implementing cognitive architectures that enable robots to build multiscale models of their users and adapt their behaviour to maximise assistance effectiveness over extended periods of interaction. I will argue that such personalisation, and the explainability of the learned user models, can help in developing effective and trustworthy robot assistants. I will illustrate our approach with examples from robots assisting people with activities of daily living, for example, dressing and mobility, as well as educational support tasks.
Yiannis Demiris is a Professor of Human-Centered Robotics at Imperial College London, where he holds the Royal Academy of Engineering Chair in Emerging Technologies. He directs the Personal Robotics Lab, which investigates multimodal perception, multiscale user modelling, and cognitive architectures, for developing effective and trustworthy robot assistants, with a particular interest in healthcare and assisted living. He has published more than 200 peer-reviewed journal and conference papers on these topics while maintaining multidisciplinary national and international collaborations with academic and commercial organisations in human-robot interaction projects, including most recently the UKRI TAS Node on Trust. He maintains a strong interest in robotics education, having received multiple teaching awards at Imperial including the Rector’s Award on Teaching Excellence. He is a Fellow of BCS, IET and RSS.
IET Public Lecture
An Uncertain Mission: Decision Making for Autonomous Robots
Nick Hawes, University of Oxford
Autonomous systems, including robots and voice assistants, are becoming increasingly capable of performing useful actions such as moving between two locations or looking up information from a website. To go beyond single actions, an autonomous system needs the ability to produce sequences of actions to allow it to achieve a user-specified goal (such as collecting then delivering a package or booking a holiday). Within the field of AI, the capability of creating these goal-directed action sequences is known as mission planning. Creating mission planning algorithms to control robots is particularly challenging because the outcomes of robot actions are often unpredictable, and may only achieve the desired outcome with some probability. In this talk, I’ll motivate the need for mission planning algorithms for robots that model uncertainty, present our approach to this problem, and talk about how we’ve used our algorithms to control robots in a range of domains from greeting people in a hospital, to inspecting a nuclear waste store.
Nick Hawes is an Associate Professor of Engineering Science in the Oxford Robotics Institute at the University of Oxford. He is also a Tutorial Fellow at Pembroke College and a Group Leader for AI/Robotics at the Turing Institute. His research is focused on developing Artificial Intelligence techniques that allow robots to perform useful tasks in everyday environments, with a particular interest in long-term autonomy, mobile service robots, and logistics. His work tackles the two connected problems of how robots can model the world around them (e.g. where objects usually appear, how people move through buildings etc.), and how they can exploit these models to perform tasks efficiently, safely, and intelligently.
Posters on Wednesday 08 September 2021 (POSTERS1)
|1.01 Reinforcement Learning-based Mapless Navigation with Fail-safe Localisation – Feiqiang Lin, et al.|
|1.02 Collaborative Coverage for a Network of Vacuum Cleaner Robots – Junyan Hu, et al.|
|1.03 Network-Aware Genetic Algorithms for the Coordination of MALE UAV Networks – Alexandros Giagkos, et al.|
|1.04 Self-organised Flocking of Robotic Swarm in Cluttered Environments – Zheyu Liu, et al.|
|1.05 Deep Learning Traversability Estimator for Mobile Robots in Unstructured Environments – Marco Visca, et al.|
|1.06 Evaluating Feedback Modalities in a Mobile Robot for Telecare – Noa Markfeld, et al.|
|1.07 Demonstrating the Differential Impact of Flock Heterogeneity on Multi-Agent Herding – Chris Bennett, et al.|
|1.08 Evaluation of an OpenCV Implementation of Structure from Motion on Open Source Data – Ali Alouache and Qinghe Wu|
|1.09 Benchmark of visual and 3D lidar SLAM systems in simulation environment for vineyards – Ibrahim Hroob, et al.|
|1.10 Lidar-only localization in 3D Pose-Feature Map – Chaokun Zhang, et al.|
|1.11 Toward robust visual odometry using prior 2D map information – S. Edwards, et al.|
|1.12 Comparison of Concentrated and Distributed Compliant Elements in a 3D Printed Gripper – Jordan Cormack, et al.|
|1.13 Perception of a humanoid robot as an interface for auditory testing – Luke Meyer, et al.|
Thursday 9 September 2021
|09:30-10:00||online system opens/networking|
|10:00-11:00||Keynote (Chair: Charles Fox)||Barbara Webb – Ant-inspired Robotics|
|11:00-11:20||UKRAS Best Paper presentation||Chipp Jansen and Elizabeth Sklar – Predicting Artist Drawing Activity via Multi-Camera Inputs for Co-Creative Drawing|
|11:20-11:45||QMUL Best thesis presentation (Chair: Lorenzo Jamone)||Shortlist: Katie Winkle, Fan Zhang, Pierre Berthet-Rayne|
|12:00-12:20||ORAL4 (Chair: Fanta Camara)||Alfred Wilmot and Ian S. Howard – 3D printed mechanically modular two-degree-of-freedom robotic segment utilizing variable-stiffness actuators|
|12:20-12:40||Oliver Shorthose, et al. – Design of a Multimaterial 3D-printed Soft Actuator with Bi-directional Variable Stiffness|
|12:40-13:00||Jack M. Frampton, et al. – Designing a Multi-Locomotion Modular Snake Robot|
|13:00-14:00||lunch/posters||Enjoy the poster showreel and some food/drink|
|14:00-14:20||ORAL5 (Chair: Athanasios Polydoros)||Marta Crivellari, et al. – Deep robot path planning from demonstrations for breast cancer examination|
|14:20-14:40||Daniel Schäle, et al. – Priors inspired by Speed-Accuracy Trade-Offs for Incremental Learning of Probabilistic Movement Primitives|
|14:40-15:00||Kiyanoush Nazari, et al. – Tactile Dynamic Behaviour Prediction Based on Robot Action|
|15:00-16:00||POSTERS2 (Chair: Marc Hanheide)||Second poster session|
|16:00-16:20||ORAL6 (Chair: Grzegorz Cielniak)||Vibhakar Mohta, et al. – Collision Avoidance with Optimal Path Replanning for Mobile Robots (swapped with first talk of the session)|
|16:20-16:40||Hatem Fakhruldeen, et al. – Development of a ROS Driver and Support Stack for the KMR iiwa Mobile Manipulator|
|16:40-17:00||Ian S. Howard and Martin F. Stoelen – State space analysis of variable-stiffness tendon drive with non-back-drivable worm-gear motor actuation (swapped with last talk of the session)|
Barbara Webb, University of Edinburgh
Ants are highly capable of many behaviours relevant to robotics. Their navigational abilities have been the focus of behavioural and ethological study for many years, and a range of algorithmic models of their behaviour have been proposed, often tested in robot implementations. Our recent work has focussed on bridging the gap to understanding the neural circuits that underly capacities such as visual orientation, path integration, and the combination of multiple cues. In each case, there is an important interplay between exploiting critical sensory cues in the natural environment, and the efficient and robust computation that supports behavioural control. A new direction for this research is to investigate the manipulation capabilities of ants, which allow them to handle a wide diversity of arbitrary, unknown objects with a skill that goes well beyond current robotics.
Barbara Webb completed a BSc in Psychology at the University of Sydney then a PhD in Artificial Intelligence at the University of Edinburgh. Her PhD research on building a robot model of cricket sound localization was featured in Scientific American and established her as a pioneer in the field of bio-robotics – using embodied models to evaluate biological hypotheses of behavioural control. She held lectureships at the University of Nottingham and University of Stirling before returning to a faculty position in the School of Informatics at Edinburgh in 2003. She was appointed to a personal chair as Professor of Biorobotics in 2010 and awarded an EPSRC Established Career Fellowship in 2021.
Posters on Thursday 09 September 2021 (POSTERS2)
|2.01 An Autonomous Mapping Approach for Confined Spaces using Flying Robots – Ahmad Alsayed, et al.|
|2.02 Maximising availability of transportation robots through intelligent allocation of parking spaces – Roopika Ravikanna, et al.|
|2.03 A Minimalist Solution to the Multi-Robot Barrier Coverage Problem – Thomas Green, et al.|
|2.04 Scheduling Multi-robot Missions with JointTasks and Heterogeneous Robot Teams – Gricel Vázquez, et al.|
|2.05 Area Coverage in Two-Dimensional Grid Worlds Using Computation-Free Agents – Arjan Dhesi and Roderich Gross|
|2.06 Online Scene Visibility Estimation as a Complement to SLAM in UAVs – Roy Haggart and Jonathan Aitken|
|2.07 Statics Optimization of a Hexapedal Robot Modelled as a Stewart Platform – Enrico Donato, et al.|
|2.08 EtherCAT implementation of a variable-stiffness tendon drive with non-back-drivable worm-gear motor actuation – Ian S. Howard and Martin F. Stoelen|
|2.09 Growing Robotic Endoscope for early Breast Cancer Detection: Robot Motion Control – Carman Larrea, et al.|
|2.10 Design and Characterisation of a Variable-Stiffness Soft Actuator Based on Tendon Twisting – William King, et al.|
|2.11 WhiskEye: A biomimetic model of multi-sensory spatial memory based on sensory reconstruction – Thomas C. Knowles, et al.|
|2.12 Equipment Detection based Inspection Robot for Industrial Plants – Mohamed Heshmat and Yang Gao|
|2.13 Inference of Mechanical Properties of Dynamic Objects through Active Perception – Nikolaus Wagner and Grzegorz Cielniak|
|2.14 Deep Learning Traversability Estimator for Mobile Robots in Unstructured Environments – Marco Visca, et al.|
- The TAROS 2021 will be virtual, oral presenters will present remotely via Zoom.
- All poster presentations will be broadcast as show reels at the start of each poster session, with opportunity to connect with presenters via Discord.
Research and Industry Showcase Day
Friday, 10 September 2021
|09:25 to 09:30||Welcome|
|10:00||GMV NSL and Sundance Ltd
Aron Kisdi and Flemming Christensen
|10:30|| University of Oxford
|11:00|| RAR UK Automation
|11:30||RAR UK Automation
|12:00||Imperial College London
|13:00 to 13:30||Lunch|
|14:30||University of Surry
|15:00||Offshore Renewable Energy Catapult
|15:30||University of Lincoln
With TAROS being fully virtual and online in 2021, we had to take the decision to change the originally planned industry and research showcase day to a virtual event as well. Consequently, we cordially invite industries and academic research labs, active in Robotics and Autonomous Systems (RAS) research, to exhibit your innovative products, research and capabilities on Friday, 10th September in short online sessions. This can be just a short presentation of your capabilities and latest exciting results, or a live tour of your lab, or a presentation of your respective product or project. We trust you will participate in this virtual exhibition and enjoy the comfort of demonstrating from your respective labs or offices. You will be given a 15-minutes slot to present yourself to the audience of TAROS, engage with the participants and increase the visibility of your robotics research and products, free of charge!
Please register directly on Eventbrite with the ticket type “Industry & Research Showcase Day Application“, which will ask you for a short description of your activity and a title so you can be featured in the programme. The programme committee will cluster the contributions thematically to put together an attractive programme. Please register your interest to be part of the showcase exhibition by Sunday,
15th 22nd (extended!) August 2021. The session will be hosted on Zoom. The programme committee reserves the right to not accommodate all requested presentations in case of oversubscription of the event or out-of-scope proposals. The programme committee will confirm your session slot by 31/8/2021.
We’ll be happy to offer you more guidance and support should you have any queries. Please contact the TAROS Industry Support Chair, Prof Mini C. Saaj, to discuss your participation (email@example.com).
Due to the still ongoing COVID-19 pandemic, we regret that TAROS 2021 again has to be purely virtual. However, this of course allows us to offer participation free of charge to anyone interested. TAROS 2021 will be hosted on Zoom, with additional interaction and socialising facilitated by the conference’s dedicated Discord server. Registration is required via Eventbrite. Participants will receive the respective links to join the conference and conversations after registration.
There are three different tickets:
- General Attendee: This is for anyone wanting to attend the conference, but not being a presenter
- Conference Presenter: If you have a paper accepted at TAROS and you will be the main presenter, please choose this option, and provide your paper number during the registration process
- Industry & Research Showcase Day: This option is for exhibitors and presenters of the TAROS Research and Industry Showcase Day only (see above). Please include a title and an abstract of your demo/exhibit/presentation.
Relevant Research Areas
Contributions are sought in all areas relevant to the overall goal of the conference, including, but not limited to the following topics:
- Autonomous robots
- Assistive, service and personal robotics
- Aerial robots, space and planetary robotics
- Medical robotics, robots for surgery and intelligent prosthetics
- Humanoid and walking robots
- Soft and flexible robots
- Swarm, collective and field robotics
- Modular reconfigurable robots
- Human-Robot Interaction
- Multimodal interfaces
- Navigation, localisation, map building and path planning,
- Analysis of robot-environment interaction
- Applications development, hardware issues, devices and techniques
- Advanced sensors and actuators
- Advanced materials and techniques
- Autonomous vehicles
- Bio-mimetic, bio-inspired and bio-hybrid robotic systems
- Evolutionary, cognitive and developmental robotics
- Robot communication and language
- Ethical and societal issues in robotics
- Robots in education, arts and entertainment
- Safety, verification and validation for robotic applications
- Learning and adaptation
- Long-term interaction and operation
- Modelling and analysis
This list of topics is not exhaustive. Due to the interdisciplinary nature of the conference, relevant contributions from other fields will also be welcome.
TAROS 2021 invites both full papers and extended abstracts. The proceedings will be published in the Springer Lecture Notes in the Artificial Intelligence series. All contributions will be refereed. Full papers are invited from researchers at any stage in their career but should present significant findings and advances in robotics research; more preliminary work would be better suited to extended abstract submission.
Full papers (up to 10 pages in LNCS format) will be accepted for either oral presentation (single track) or poster presentation. Extended abstracts (up to 4 pages in LNCS format) will be accepted for poster presentation only.
Submissions MUST be in LNCS format, templates are available from:
The submission system is now closed, authors can still log in to access their submissions at https://easychair.org/conferences/?conf=taros2021.
Prof. Mini C. Saaj