A5030073861- Reinforcement Learning in Robotics
- Robotic Locomotion and Control
- Human Pose and Action Recognition
- Robot Manipulation and Learning
- Adversarial Robustness in Machine Learning
- Prosthetics and Rehabilitation Robotics
- Real-time simulation and control systems
- Machine Learning and Algorithms
- Advanced Adaptive Filtering Techniques
- Blind Source Separation Techniques
- Magnetic Bearings and Levitation Dynamics
- Sports Analytics and Performance
- Advanced Vision and Imaging
- Domain Adaptation and Few-Shot Learning
- Speech and Audio Processing
- Winter Sports Injuries and Performance
- Sensorless Control of Electric Motors
- Muscle activation and electromyography studies
- Terahertz technology and applications
- Advanced Neural Network Applications
- Explainable Artificial Intelligence (XAI)
- Superconducting and THz Device Technology
- Electric Motor Design and Analysis
- Model Reduction and Neural Networks
- Advanced Memory and Neural Computing
DeepMind (United Kingdom)
2022-2024
Google (United Kingdom)
2024
University College London
2023
University of California, Berkeley
2016-2020
Google (United States)
2019
Machine Intelligence Research Institute
2019
Intel (United States)
2018
VTT Technical Research Centre of Finland
2008-2014
Aalto University
2010-2011
Model-free deep reinforcement learning (RL) algorithms have been demonstrated on a range of challenging decision making and control tasks. However, these methods typically suffer from two major challenges: very high sample complexity brittle convergence properties, which necessitate meticulous hyperparameter tuning. Both challenges severely limit the applicability such to complex, real-world domains. In this paper, we propose soft actor-critic, an off-policy actor-critic RL algorithm based...
Model-free deep reinforcement learning (RL) algorithms have been successfully applied to a range of challenging sequential decision making and control tasks. However, these methods typically suffer from two major challenges: high sample complexity brittleness hyperparameters. Both challenges limit the applicability such real-world domains. In this paper, we describe Soft Actor-Critic (SAC), our recently introduced off-policy actor-critic algorithm based on maximum entropy RL framework....
Deep reinforcement learning (deep RL) holds the promise of automating acquisition complex controllers that can map sensory inputs directly to low-level actions.In domain robotic locomotion, deep RL could enable locomotion skills with minimal engineering and without an explicit model robot dynamics.Unfortunately, applying real-world tasks is exceptionally difficult, primarily due poor sample complexity sensitivity hyperparameters.While hyperparameters be easily tuned in simulated domains,...
Model-free deep reinforcement learning has been shown to exhibit good performance in domains ranging from video games simulated robotic manipulation and locomotion. However, model-free methods are known perform poorly when the interaction time with environment is limited, as case for most real-world tasks. In this paper, we study how maximum entropy policies trained using soft Q-learning can be applied manipulation. The application of method facilitated by two important features Q-learning....
We investigated whether deep reinforcement learning (deep RL) is able to synthesize sophisticated and safe movement skills for a low-cost, miniature humanoid robot that can be composed into complex behavioral strategies. used RL train play simplified one-versus-one soccer game. The resulting agent exhibits robust dynamic skills, such as rapid fall recovery, walking, turning, kicking, it transitions between them in smooth efficient manner. It also learned anticipate ball movements block...
Generative state estimators based on probabilistic filters and smoothers are one of the most popular classes for robots autonomous vehicles. However, generative models have limited capacity to handle rich sensory observations, such as camera images, since they must model entire distribution over sensor readings. Discriminative do not suffer from this limitation, but typically more complex train latent variable estimation. We present an alternative approach where parameters directly optimized...
Learning to combine control at the level of joint torques with longer-term goal-directed behavior is a long-standing challenge for physically embodied artificial agents. Intelligent in physical world unfolds across multiple spatial and temporal scales: Although movements are ultimately executed instantaneous muscle tensions or torques, they must be selected serve goals that defined on much longer time scales often involve complex interactions environment other Recent research has...
We address the problem of learning hierarchical deep neural network policies for reinforcement learning. In contrast to methods that explicitly restrict or cripple lower layers a hierarchy force them use higher-level modulating signals, each layer in our framework is trained directly solve task, but acquires range diverse strategies via maximum entropy objective. Each also augmented with latent random variables, which are sampled from prior distribution during training layer. The objective...
We present a system for applying sim2real approaches to "in the wild" scenes with realistic visuals, and policies which rely on active perception using RGB cameras. Given short video of static scene collected generic phone, we learn scene's contact geometry function novel view synthesis Neural Radiance Field (NeRF). augment NeRF rendering by overlaying other dynamic objects (e.g. robot's own body, ball). A simulation is then created engine in physics simulator computes dynamics from...
We investigate the use of prior knowledge human and animal movement to learn reusable locomotion skills for real legged robots. Our approach builds upon previous work on imitating or dog Motion Capture (MoCap) data a skill module. Once learned, this module can be reused complex downstream tasks. Importantly, due imposed by MoCap data, our does not require extensive reward engineering produce sensible natural looking behavior at time reuse. This makes it easy create well-regularized,...
Deep reinforcement learning (deep RL) holds the promise of automating acquisition complex controllers that can map sensory inputs directly to low-level actions. In domain robotic locomotion, deep RL could enable locomotion skills with minimal engineering and without an explicit model robot dynamics. Unfortunately, applying real-world tasks is exceptionally difficult, primarily due poor sample complexity sensitivity hyperparameters. While hyperparameters be easily tuned in simulated domains,...
Intelligent behaviour in the physical world exhibits structure at multiple spatial and temporal scales. Although movements are ultimately executed level of instantaneous muscle tensions or joint torques, they must be selected to serve goals defined on much longer timescales, terms relations that extend far beyond body itself, involving coordination with other agents. Recent research artificial intelligence has shown promise learning-based approaches respective problems complex movement,...
Passive imaging of concealed objects at stand-off distances in excess a few meters requires both excellent spatial, thermal and temporal resolution from the terahertz system. The combination these requirements while keeping overall system cost reasonable level has been motivation for this joint work. THz under development is capable sub-Kelvin NETD video frame rates. In paper we report first results 16-pixel array superconducting antenna-coupled NbN vacuum-bridge microbolometers, operated...
Reinforcement learning requires manual specification of a reward function to learn task. While in principle this only needs specify the task goal, practice reinforcement can be very time-consuming or even infeasible unless is shaped so as provide smooth gradient towards successful outcome. This shaping difficult by hand, particularly when learned from raw observations, such images. In paper, we study how automatically dynamical distances: measure expected number time steps reach given goal...
This paper shows a general review of Switched Reluctance Motors (SRM), reports the conversion Active Magnetic Bearing (AMB) to SRM and discusses methods estimate torque produced in SRM. AMB converted has been tested on laboratory scale test rig its production measured. The novelty this is modified analytical method fringing flux paths enabling more accurate mathematical model torque. reason for careful analysis air gap forces demand electromagnetic models motor be used controller. ultimate...