Student presentation – Alex Hawkins Hooker at ISMB
By sharon.betts, on 4 October 2023
UKRI Centre for Doctoral Training in Foundational AI
HomeBy sharon.betts, on 4 October 2023
By sharon.betts, on 29 September 2023
In the exciting realm of Model-based Reinforcement Learning (MBRL), researchers are constantly pushing the boundaries of what robots can learn to achieve when given access to an internal model of the environment. One key challenge in this field is ensuring that robots can perform tasks safely and reliably, especially in situations where they lack prior data or knowledge about the environment. That’s where the work of Sicelukwanda Zwane comes into play.
Background
In MBRL, robots use small sets of data to learn a dynamics model. This model is like a crystal ball that predicts how the system will respond to a given sequence of different actions. With MBRL, we can train policies from simulated trajectories sampled from the dynamics model instead of first generating them by executing each action on the actual system, a process that can take extremely long periods of time on a physical robot and possibly cause wear and tear.
One of the tools often used in MBRL is the Gaussian process (GP) dynamics model. GPs are fully-Bayesian models that not only model the system but also account for the uncertainty in state observations. Additionally, they are flexible and are able to learn without making strong assumptions about the underlying system dynamics [1].
The Challenge of Learning Safely
When we train robots to perform tasks, it’s not enough to just predict what will happen; we need to do it safely. As with most model classes in MBRL, GPs don’t naturally incorporate safety constraints. This means that they may produce unsafe or unfeasible trajectories. This is particularly true during early stages of learning, when the model hasn’t seen much data, it can produce unsafe and seemingly random trajectories.
For a 7 degree of freedom (DOF) manipulator robot, bad trajectories may contain self-collisions.
Distributional Trajectory Sampling
In standard GP dynamics models, the posterior is represented in distributional form – using its parameters, the mean vector and covariance matrix. In this form, it is difficult to reason about
about the safety of entire trajectories. This is because trajectories are generated through iterative random sampling. Furthermore, this kind of trajectory sampling is limited to cases where the intermediate state marginal distributions are Gaussian distributed.
Pathwise Trajectory Sampling
Zwane uses an innovative alternative called “pathwise sampling” [3]. This approach draws samples from GP posteriors using an efficient method called Matheron’s rule. The result is a set of smooth, deterministic trajectories that aren’t confined to Gaussian distributions and are temporally correlated.
Adding Safety
The beauty of pathwise sampling [3] is that it has a particle representation of the GP posterior, where individual trajectories are smooth, differentiable, and deterministic functions. This allows for the isolation of constraint-violating trajectories from safe ones. For safety, rejection sampling is performed on trajectories that violate safety constraints, leaving behind only the safe ones to train the policy. Additionally, soft constraint penalty terms are added to the reward function.
Sim-Real Robot Experiments
To put this approach to the test, Zwane conducted experiments involving a 7-DoF robot arm in a simulated constrained reaching task, where the robot has to avoid colliding with a low ceiling. The method successfully learned a reaching policy that adhered to safety constraints, even when starting from random initial states.
In this constrained manipulation task, the robot is able to reach the goal (shown by the red sphere – bottom row) without colliding with the ceiling (blue – bottom row) using less than 100 seconds of data in simulation.
Summary
Sicelukwanda Zwane’s research makes incremental advances on the safety of simulated trajectories by incorporating safety constraints while keeping the benefits of using fully-Bayesian dynamics models such as GPs. This method promises to take MBRL out of simulated environments and make it more applicable to real-world settings. If you’re interested in this work, we invite you to dive into the full paper, published at the recent IEEE CASE 2023 conference.
References
By sharon.betts, on 29 November 2022
On 7th and 8th November 2022 three of the UKRI CDTs in Artificial Intelligence hosted an Inter-CDT conference for our students and industry partners at The Bristol Hotel. The UKRI CDT in Foundational AI worked alongside our sister CDTs at the University of Bath (ART-AI) and University of Bristol (Interactive AI), to produce a two day event that covered AI from deep tech entrepreneurship to AI Ethics and Defence.
Turnout from all three CDTs was excellent and it was a wonderful opportunity for students across the three institutions to meet and collaborate with one another, sharing their knowledge and research of AI both in theory and applied.
UCL were delighted to host two panel sessions; the first being on Deep Tech entrepreneurship with Dr. Riam Kanso from Conception X, Dr. Stacy-Ann Sinclair from CodeREG and Dr. Thomas Stone from Kintsugi (ad)Ventures. Hosted by our CDT Director, Prof David Barber, this interactive panel session saw our specialists discuss the pathways into start ups and entrepreneurships, the perils, pitfalls and positives that follow! It was wonderful to be able to hear from industry experts their personal journeys to successful business ventures and great to have such an engaged and enquiring audience, who were keen to ask numerous questions and gain further insight to future possibilities.
Our second panel closed the event and was a student-led initiative discussing large scale datasets and massive computational modelling in AI.
For a more detailed review of the event we highly recommend you read the review by ART-AI on their website.
We were delighted to celebrate our student Dennis Hadjivelichkov’s second place in the poster session that took place at the MShed in Bristol as well as enjoy the fine food and fabulous company of our CDT peers.
With thanks to ART-AI and Interactive AI CDTs for their co-hosting and co-organising skills. It was a delight to be able to share time and work with our sister CDTs and we hope to collaborate again in the not too distant future.
By sharon.betts, on 16 November 2022
In September 2022 I had the amazing opportunity to participate in workshops in Rennes and in Sophia Antipolis, and in a doctoral symposium in Alicante!
In poster sessions and talks, I have presented my work on Aggregated Kernel Tests which covers three of my papers. The first one is MMD Aggregated Two-Sample Test where the two-sample problem is considered, in which one has access to samples from two distributions and is interested in detecting whether those come from the same or from different distributions. The second is KSD Aggregated Goodness-of-fit Test in which we consider the goodness-of-fit problem where one is given some samples and is asked whether these come from a given model (with access to its density or score function). In the third one, Efficient Aggregated Kernel Tests using Incomplete U-statistics, we propose computationally efficient tests for the two-sample, goodness-of-fit, and independence problems; this last one consists in detecting dependence between the two components of paired samples. We tackle these three testing problems using kernel-based statistics, in such a setting the performance of these tests is known to heavily depend on the choice of kernels or kernel parameters (i.e. bandwidth parameter). We propose tests which aggregate over a collection of kernels and retain test power, we theoretically prove optimality of our tests under some regularity assumptions, and empirically show that our aggregated tests outperform other state-of-the-art kernel-based tests.
I am extremely grateful to Valentin Patilea, Motonobu Kanagawa and Aditya Gulati for the respective invitations, and to my CDT (UCL CDT in Foundational AI with funding from UKRI) which allowed me to participate in those workshops/symposium!