Language-Informed Transfer Learning for Embodied Household Activities.
Jiang, Yuqian; Gao, Qiaozi; Thattai, Govind; Sukhatme, Gaurav S.: Language-Informed Transfer Learning for Embodied Household Activities. CoRR, vol. abs/2301.05318, 2023
Publications
RREx-BoT: Remote Referring Expressions with a Bag of Tricks.
Sigurdsson, Gunnar A.; Thomason, Jesse; Sukhatme, Gaurav S.; Piramuthu, Robinson: RREx-BoT: Remote Referring Expressions with a Bag of Tricks. CoRR, vol. abs/2301.12614, 2023
[Blog]
[Website]
Learning Robot Manipulation from Cross-Morphology Demonstration
Salhotra, Gautam; Liu, I-Chun Arthur; Sukhatme, Gaurav S.: Learning Robot Manipulation from Cross-Morphology Demonstration Conference on Robot Learning (CoRL) 2023, 2023
Proximal Policy Gradient Arborescence for Quality Diversity Reinforcement Learning.
Batra, Sumeet; Tjanaka, Bryon; Fontaine, Matthew C.; Petrenko, Aleksei; Nikolaidis, Stefanos; Sukhatme, Gaurav S.: Proximal Policy Gradient Arborescence for Quality Diversity Reinforcement Learning. Submitted to Neurips 2023, 2023
[Twitter]
Generating Behaviorally Diverse Policies with Latent Diffusion Models.
Hegde, Shashank; Batra, Sumeet; Zentner, K. R.; Sukhatme, Gaurav S.: Generating Behaviorally Diverse Policies with Latent Diffusion Models. Submitted to Neurips 2023, 2023
[Twitter]
[Website]
CH-MARL: A Multimodal Benchmark for Cooperative, Heterogeneous Multi-Agent Reinforcement Learning.
Sharma, Vasu; Goyal, Prasoon; Lin, Kaixiang; Thattai, Govind; Gao, Qiaozi; Sukhatme, Gaurav S.: CH-MARL: A Multimodal Benchmark for Cooperative, Heterogeneous Multi-Agent Reinforcement Learning. CoRR, vol. abs/2208.13626, 2022
Efficiently Learning Small Policies for Locomotion and Manipulation.
Hegde, Shashank; Sukhatme, Gaurav S.: Efficiently Learning Small Policies for Locomotion and Manipulation. CoRR, vol. abs/2210.00140, 2022
[Video]
CLIP-Nav: Using CLIP for Zero-Shot Vision-and-Language Navigation.
Dorbala, Vishnu Sashank; Sigurdsson, Gunnar A.; Piramuthu, Robinson; Thomason, Jesse; Sukhatme, Gaurav S.: CLIP-Nav: Using CLIP for Zero-Shot Vision-and-Language Navigation. CoRR, vol. abs/2211.16649, 2022
Efficient Multi-Task Learning via Iterated Single-Task Transfer.
Zentner, K. R.; Puri, Ujjwal; Zhang, Yulun; Julian, Ryan; Sukhatme, Gaurav S.: Efficient Multi-Task Learning via Iterated Single-Task Transfer. IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2022, Kyoto, Japan, October 23-27, 2022, pp. 10141-10146, IEEE, 2022
Supervised learning and reinforcement learning of feedback models for reactive behaviors: Tactile feedback testbed.
Sutanto, Giovanni; Rombach, Katharina; Chebotar, Yevgen; Su, Zhe; Schaal, Stefan; Sukhatme, Gaurav S.; Meier, Franziska: Supervised learning and reinforcement learning of feedback models for reactive behaviors: Tactile feedback testbed. Int. J. Robotics Res., vol. 41, no. 13-14, pp. 1121-1145, 2022
Suboptimal coverings for continuous spaces of control tasks.
Preiss, James A.; Sukhatme, Gaurav S.: Suboptimal coverings for continuous spaces of control tasks. CoRR, vol. abs/2104.11865, 2021
Towards Exploiting Geometry and Time for Fast Off-Distribution Adaptation in Multi-Task Robot Learning.
Zentner, K. R.; Julian, Ryan; Puri, Ujjwal; Zhang, Yulun; Sukhatme, Gaurav S.: Towards Exploiting Geometry and Time for Fast Off-Distribution Adaptation in Multi-Task Robot Learning. CoRR, vol. abs/2106.13237, 2021
Decentralized Control of Quadrotor Swarms with End-to-end Deep Reinforcement Learning.
Batra, Sumeet; Huang, Zhehui; Petrenko, Aleksei; Kumar, Tushar; Molchanov, Artem; Sukhatme, Gaurav S.: Decentralized Control of Quadrotor Swarms with End-to-end Deep Reinforcement Learning. CoRR, vol. abs/2109.07735, 2021
Adaptive Sampling using POMDPs with Domain-Specific Considerations.
Salhotra, Gautam; Denniston, Christopher E.; Caron, David A.; Sukhatme, Gaurav S.: Adaptive Sampling using POMDPs with Domain-Specific Considerations. International Conference on Robotics and Automation (ICRA 2020), 2021
Beyond Robustness: A Taxonomy of Approaches towards Resilient Multi-Robot Systems.
Prorok, Amanda; Malencia, Matthew; Carlone, Luca; Sukhatme, Gaurav S.; Sadler, Brian M.; Kumar, Vijay: Beyond Robustness: A Taxonomy of Approaches towards Resilient Multi-Robot Systems. CoRR, vol. abs/2109.12343, 2021
A Simple Approach to Continual Learning by Transferring Skill Parameters.
Zentner, K. R.; Julian, Ryan; Puri, Ujjwal; Zhang, Yulun; Sukhatme, Gaurav S.: A Simple Approach to Continual Learning by Transferring Skill Parameters. CoRR, vol. abs/2110.10255, 2021
From Machine Learning to Robotics: Challenges and Opportunities for Embodied Intelligence.
Roy, Nicholas; Posner, Ingmar; Barfoot, Tim D.; Beaudoin, Philippe; Bengio, Yoshua; Bohg, Jeannette; Brock, Oliver; Depatie, Isabelle; Fox, Dieter; Koditschek, Daniel E.; Lozano-PØrez, TomÆs; Mansinghka, Vikash; Pal, Christopher J.; Richards, Blake A.; Sadigh, Dorsa; Schaal, Stefan; Sukhatme, Gaurav S.; ThØrien, Denis; Toussaint, Marc; Panne, Michiel: From Machine Learning to Robotics: Challenges and Opportunities for Embodied Intelligence. CoRR, vol. abs/2110.15245, 2021
LUMINOUS: Indoor Scene Generation for Embodied AI Challenges.
Zhao, Yizhou; Lin, Kaixiang; Jia, Zhiwei; Gao, Qiaozi; Thattai, Govind; Thomason, Jesse; Sukhatme, Gaurav S.: LUMINOUS: Indoor Scene Generation for Embodied AI Challenges. CoRR, vol. abs/2111.05527, 2021
Decentralized Control of Quadrotor Swarms with End-to-end Deep Reinforcement Learning.
Batra, Sumeet; Huang, Zhehui; Petrenko, Aleksei; Kumar, Tushar; Molchanov, Artem; Sukhatme, Gaurav S.: Decentralized Control of Quadrotor Swarms with End-to-end Deep Reinforcement Learning. Faust, Aleksandra; Hsu, David; Neumann, Gerhard (Ed.): Conference on Robot Learning, 8-11 November 2021, London, UK, pp. 576-586, PMLR, 2021
Adaptive Sampling using POMDPs with Domain-Specific Considerations.
Salhotra, Gautam; Denniston, Christopher E.; Caron, David A.; Sukhatme, Gaurav S.: Adaptive Sampling using POMDPs with Domain-Specific Considerations. IEEE International Conference on Robotics and Automation, ICRA 2021, Xi'an, China, May 30 - June 5, 2021, pp. 2385-2391, IEEE, 2021
NeuralSim: Augmenting Differentiable Simulators with Neural Networks.
Heiden, Eric; Millard, David; Coumans, Erwin; Sheng, Yizhou; Sukhatme, Gaurav S.: NeuralSim: Augmenting Differentiable Simulators with Neural Networks. IEEE International Conference on Robotics and Automation, ICRA 2021, Xi'an, China, May 30 - June 5, 2021, pp. 9474-9481, IEEE, 2021
[Video]
Suboptimal coverings for continuous spaces of control tasks.
Preiss, James A.; Sukhatme, Gaurav S.: Suboptimal coverings for continuous spaces of control tasks. Jadbabaie, Ali; Lygeros, John; Pappas, George J.; Parrilo, Pablo A.; Recht, Benjamin; Tomlin, Claire J.; Zeilinger, Melanie N. (Ed.): Proceedings of the 3rd Annual Conference on Learning for Dynamics and Control, L4DC 2021, 7-8 June 2021, Virtual Event, Switzerland, pp. 547-558, PMLR, 2021
Meta Learning via Learned Loss.
Bechtle, Sarah; Molchanov, Artem; Chebotar, Yevgen; Grefenstette, Edward; Righetti, Ludovic; Sukhatme, Gaurav S.; Meier, Franziska: Meta Learning via Learned Loss. 25th International Conference on Pattern Recognition, ICPR 2020, Virtual Event / Milan, Italy, January 10-15, 2021, pp. 4161-4168, IEEE, 2020
Encoding Physical Constraints in Differentiable Newton-Euler Algorithm.
Sutanto, Giovanni; Wang, Austin S.; Lin, Yixin; Mukadam, Mustafa; Sukhatme, Gaurav S.; Rai, Akshara; Meier, Franziska: Encoding Physical Constraints in Differentiable Newton-Euler Algorithm. Bayen, Alexandre M.; Jadbabaie, Ali; Pappas, George J.; Parrilo, Pablo A.; Recht, Benjamin; Tomlin, Claire J.; Zeilinger, Melanie N. (Ed.): Proceedings of the 2nd Annual Conference on Learning for Dynamics and Control, L4DC 2020, Online Event, Berkeley, CA, USA, 11-12 June 2020, pp. 804-813, PMLR, 2020
Encoding Physical Constraints in Differentiable Newton-Euler Algorithm.
Sutanto, Giovanni; Wang, Austin S.; Lin, Yixin; Mukadam, Mustafa; Sukhatme, Gaurav S.; Rai, Akshara; Meier, Franziska: Encoding Physical Constraints in Differentiable Newton-Euler Algorithm. CoRR, vol. abs/2001.08861, 2020
Efficient Adaptation for End-to-End Vision-Based Robotic Manipulation.
Julian, Ryan; Swanson, Benjamin; Sukhatme, Gaurav S.; Levine, Sergey; Finn, Chelsea; Hausman, Karol: Efficient Adaptation for End-to-End Vision-Based Robotic Manipulation. CoRR, vol. abs/2004.10190, 2020
Supervised Learning and Reinforcement Learning of Feedback Models for Reactive Behaviors: Tactile Feedback Testbed.
Sutanto, Giovanni; Rombach, Katharina; Chebotar, Yevgen; Su, Zhe; Schaal, Stefan; Sukhatme, Gaurav S.; Meier, Franziska: Supervised Learning and Reinforcement Learning of Feedback Models for Reactive Behaviors: Tactile Feedback Testbed. CoRR, vol. abs/2007.00450, 2020
NeuralSim: Augmenting Differentiable Simulators with Neural Networks.
Heiden, Eric; Millard, David; Coumans, Erwin; Sheng, Yizhou; Sukhatme, Gaurav S.: NeuralSim: Augmenting Differentiable Simulators with Neural Networks. CoRR, vol. abs/2011.04217, 2020
Scaling simulation-to-real transfer by learning a latent space of robot skills.
Julian, Ryan; Heiden, Eric; He, Zhanpeng; Zhang, Hejia; Schaal, Stefan; Lim, Joseph J.; Sukhatme, Gaurav S.; Hausman, Karol: Scaling simulation-to-real transfer by learning a latent space of robot skills. Int. J. Robotics Res., vol. 39, no. 10-11, 2020
Never Stop Learning: The Effectiveness of Fine-Tuning in Robotic Reinforcement Learning.
Julian, Ryan; Swanson, Benjamin; Sukhatme, Gaurav S.; Levine, Sergey; Finn, Chelsea; Hausman, Karol: Never Stop Learning: The Effectiveness of Fine-Tuning in Robotic Reinforcement Learning. Kober, Jens; Ramos, Fabio; Tomlin, Claire J. (Ed.): 4th Conference on Robot Learning, CoRL 2020, 16-18 November 2020, Virtual Event / Cambridge, MA, USA, pp. 2120-2136, PMLR, 2020
Sample Factory: Egocentric 3D Control from Pixels at 100000 FPS with Asynchronous Reinforcement Learning.
Petrenko, Aleksei; Huang, Zhehui; Kumar, Tushar; Sukhatme, Gaurav S.; Koltun, Vladlen: Sample Factory: Egocentric 3D Control from Pixels at 100000 FPS with Asynchronous Reinforcement Learning. Proceedings of the 37th International Conference on Machine Learning, ICML 2020, 13-18 July 2020, Virtual Event, pp. 7652-7662, PMLR, 2020
Incorporating Noise into Adaptive Sampling.
Denniston, Christopher E.; Kumaraguru, Aravind; Caron, David A.; Sukhatme, Gaurav S.: Incorporating Noise into Adaptive Sampling. Siciliano, Bruno; Laschi, Cecilia; Khatib, Oussama (Ed.): Experimental Robotics - The 17th International Symposium, ISER 2020, La Valletta, Malta, November 9-12, 2020 (postponed to 2021), pp. 198-208, Springer, 2020
Estimating Metric Scale Visual Odometry from Videos using 3D Convolutional Networks.
Koumis, Alexander S.; Preiss, James A.; Sukhatme, Gaurav S.: Estimating Metric Scale Visual Odometry from Videos using 3D Convolutional Networks. 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2019, Macau, SAR, China, November 3-8, 2019, pp. 265-272, IEEE, 2019
Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors.
Molchanov, Artem; Chen, Tao; Hünig, Wolfgang; Preiss, James A.; Ayanian, Nora; Sukhatme, Gaurav S.: Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors. 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2019, Macau, SAR, China, November 3-8, 2019, pp. 59-66, IEEE, 2019
Accelerating Goal-Directed Reinforcement Learning by Model Characterization.
Debnath, Shoubhik; Sukhatme, Gaurav S.; Liu, Lantao: Accelerating Goal-Directed Reinforcement Learning by Model Characterization. CoRR, vol. abs/1901.01977, 2019
Meta-Learning via Learned Loss.
Chebotar, Yevgen; Molchanov, Artem; Bechtle, Sarah; Righetti, Ludovic; Meier, Franziska; Sukhatme, Gaurav S.: Meta-Learning via Learned Loss. CoRR, vol. abs/1906.05374, 2019
Profit Maximizing Logistic Regression Modeling for Credit Scoring.
Devos, Arnout; Dhondt, Jakob; Stripling, Eugen; Baesens, Bart; Broucke, Seppe; Sukhatme, Gaurav S.: Profit Maximizing Logistic Regression Modeling for Credit Scoring. 2018 IEEE Data Science Workshop, DSW 2018, Lausanne, Switzerland, June 4-6, 2018, pp. 125-129, IEEE, 2018
Accelerating Goal-Directed Reinforcement Learning by Model Characterization.
Debnath, Shoubhik; Sukhatme, Gaurav S.; Liu, Lantao: Accelerating Goal-Directed Reinforcement Learning by Model Characterization. 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2018, Madrid, Spain, October 1-5, 2018, pp. 1-9, IEEE, 2018
Region Growing Curriculum Generation for Reinforcement Learning.
Molchanov, Artem; Hausman, Karol; Birchfield, Stan; Sukhatme, Gaurav S.: Region Growing Curriculum Generation for Reinforcement Learning. CoRR, vol. abs/1807.01425, 2018
Zero-Shot Skill Composition and Simulation-to-Real Transfer by Learning Task Representations.
He, Zhanpeng; Julian, Ryan; Heiden, Eric; Zhang, Hejia; Schaal, Stefan; Lim, Joseph J.; Sukhatme, Gaurav S.; Hausman, Karol: Zero-Shot Skill Composition and Simulation-to-Real Transfer by Learning Task Representations. CoRR, vol. abs/1810.02422, 2018
Data-driven learning and planning for environmental sampling.
Ma, Kai-Chieh; Liu, Lantao; Heidarsson, Hordur Kristinn; Sukhatme, Gaurav S.: Data-driven learning and planning for environmental sampling. J. Field Robotics, vol. 35, no. 5, pp. 643-661, 2018
Scaling Simulation-to-Real Transfer by Learning Composable Robot Skills.
Julian, Ryan; Heiden, Eric; He, Zhanpeng; Zhang, Hejia; Schaal, Stefan; Lim, Joseph J.; Sukhatme, Gaurav S.; Hausman, Karol: Scaling Simulation-to-Real Transfer by Learning Composable Robot Skills. Xiao, Jing; Krüger, Torsten; Khatib, Oussama (Ed.): Proceedings of the 2018 International Symposium on Experimental Robotics, ISER 2018, Buenos Aires, Argentina, November 5-8, 2018, pp. 267-279, Springer, 2018
Informative planning and online learning with sparse Gaussian processes.
Ma, Kai-Chieh; Liu, Lantao; Sukhatme, Gaurav S.: Informative planning and online learning with sparse Gaussian processes. 2017 IEEE International Conference on Robotics and Automation, ICRA 2017, Singapore, Singapore, May 29 - June 3, 2017, pp. 4292-4298, IEEE, 2017
Feature selection for learning versatile manipulation skills based on observed and desired trajectories.
Kroemer, Oliver; Sukhatme, Gaurav S.: Feature selection for learning versatile manipulation skills based on observed and desired trajectories. 2017 IEEE International Conference on Robotics and Automation, ICRA 2017, Singapore, Singapore, May 29 - June 3, 2017, pp. 4713-4720, IEEE, 2017
Data-Driven Learning and Planning for Environmental Sampling.
Ma, Kai-Chieh; Liu, Lantao; Heidarsson, Hordur Kristinn; Sukhatme, Gaurav S.: Data-Driven Learning and Planning for Environmental Sampling. CoRR, vol. abs/1702.01848, 2017
Multi-Modal Imitation Learning from Unstructured Demonstrations using Generative Adversarial Nets.
Hausman, Karol; Chebotar, Yevgen; Schaal, Stefan; Sukhatme, Gaurav S.; Lim, Joseph J.: Multi-Modal Imitation Learning from Unstructured Demonstrations using Generative Adversarial Nets. Guyon, Isabelle; Luxburg, Ulrike; Bengio, Samy; Wallach, Hanna M.; Fergus, Rob; Vishwanathan, S. V. N.; Garnett, Roman (Ed.): Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems 2017, December 4-9, 2017, Long Beach, CA, USA, pp. 1235-1245, 2017
Learning spatial preconditions of manipulation skills using random forests.
Kroemer, Oliver; Sukhatme, Gaurav S.: Learning spatial preconditions of manipulation skills using random forests. 16th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2016, Cancun, Mexico, November 15-17, 2016, pp. 676-683, IEEE, 2016
Self-supervised regrasping using spatio-temporal tactile features and reinforcement learning.
Chebotar, Yevgen; Hausman, Karol; Su, Zhe; Sukhatme, Gaurav S.; Schaal, Stefan: Self-supervised regrasping using spatio-temporal tactile features and reinforcement learning. 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2016, Daejeon, South Korea, October 9-14, 2016, pp. 1960-1966, IEEE, 2016
Online trajectory optimization to improve object recognition.
Potthast, Christian; Sukhatme, Gaurav S.: Online trajectory optimization to improve object recognition. 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS 2016, Daejeon, South Korea, October 9-14, 2016, pp. 4765-4772, IEEE, 2016
Learning Relevant Features for Manipulation Skills using Meta-Level Priors.
Kroemer, Oliver; Sukhatme, Gaurav S.: Learning Relevant Features for Manipulation Skills using Meta-Level Priors. CoRR, vol. abs/1605.04439, 2016
Informative Planning and Online Learning with Sparse Gaussian Processes.
Ma, Kai-Chieh; Liu, Lantao; Sukhatme, Gaurav S.: Informative Planning and Online Learning with Sparse Gaussian Processes. CoRR, vol. abs/1609.07560, 2016
Meta-level Priors for Learning Manipulation Skills with Sparse Features.
Kroemer, Oliver; Sukhatme, Gaurav S.: Meta-level Priors for Learning Manipulation Skills with Sparse Features. Kulic, Dana; Nakamura, Yoshihiko; Khatib, Oussama; Venture, Gentiane (Ed.): International Symposium on Experimental Robotics, ISER 2016, Tokyo, Japan, October 3-6, 2016, pp. 211-222, Springer, 2016
Learning to Switch Between Sensorimotor Primitives Using Multimodal Haptic Signals.
Su, Zhe; Kroemer, Oliver; Loeb, Gerald E.; Sukhatme, Gaurav S.; Schaal, Stefan: Learning to Switch Between Sensorimotor Primitives Using Multimodal Haptic Signals. Tuci, Elio; Giagkos, Alexandros; Wilson, Myra S.; Hallam, John (Ed.): From Animals to Animats 14 - 14th International Conference on Simulation of Adaptive Behavior, SAB 2016, Aberystwyth, UK, August 23-26, 2016, Proceedings, pp. 170-182, Springer, 2016
Data-driven robotic sampling for marine ecosystem monitoring.
Das, Jnaneshwar; Py, FrØdØric; Harvey, Julio B. J.; Ryan, John P.; Gellene, Alyssa; Graham, Rishi; Caron, David A.; Rajan, Kanna; Sukhatme, Gaurav S.: Data-driven robotic sampling for marine ecosystem monitoring. Int. J. Robotics Res., vol. 34, no. 12, pp. 1435-1452, 2015
Learning task error models for manipulation.
Pastor, Peter; Kalakrishnan, Mrinal; Binney, Jonathan; Kelly, Jonathan; Righetti, Ludovic; Sukhatme, Gaurav S.; Schaal, Stefan: Learning task error models for manipulation. 2013 IEEE International Conference on Robotics and Automation, Karlsruhe, Germany, May 6-10, 2013, pp. 2612-2618, IEEE, 2013
Hierarchical probabilistic regression for AUV-based adaptive sampling of marine phenomena.
Das, Jnaneshwar; Harvey, Julio B. J.; Py, Frederic; Vathsangam, Harshvardhan; Graham, Rishi; Rajan, Kanna; Sukhatme, Gaurav S.: Hierarchical probabilistic regression for AUV-based adaptive sampling of marine phenomena. 2013 IEEE International Conference on Robotics and Automation, Karlsruhe, Germany, May 6-10, 2013, pp. 5571-5578, IEEE, 2013
Active planning for underwater inspection and the benefit of adaptivity.
Hollinger, Geoffrey A.; Englot, Brendan J.; Hover, Franz S.; Mitra, Urbashi; Sukhatme, Gaurav S.: Active planning for underwater inspection and the benefit of adaptivity. Int. J. Robotics Res., vol. 32, no. 1, pp. 3-18, 2013
Architecture-driven self-adaptation and self-management in robotics systems.
Edwards, George; Garcia, Joshua; Tajalli, Hossein; Popescu, Daniel; Medvidovic, Nenad; Sukhatme, Gaurav S.; Petrus, Brad: Architecture-driven self-adaptation and self-management in robotics systems. 2009 ICSE Workshop on Software Engineering for Adaptive and Self-Managing Systems, SEAMS 2009, Vancouver, BC, Canada, May 18-19, 2009, pp. 142-151, IEEE Computer Society, 2009
Reconfiguration methods for mobile sensor networks.
Kansal, Aman; Kaiser, William J.; Pottie, Gregory J.; Srivastava, Mani B.; Sukhatme, Gaurav S.: Reconfiguration methods for mobile sensor networks. ACM Trans. Sens. Networks, vol. 3, no. 4, pp. 22, 2007
The 2002 AAAI Spring Symposium Series.
Karlgren, Jussi; Kanerva, Pentti; Gambîck, Bjürn; Forbus, Kenneth D.; Tumer, Kagan; Stone, Peter; Goebel, Kai; Sukhatme, Gaurav S.; Balch, Tucker R.; Fischer, Bernd; Smith, Doug; Harabagiu, Sanda M.; Chaudri, Vinay K.; Barley, Mike; Guesgen, Hans W.; Stahovich, Thomas F.; Davis, Randall; Landay, James A.: The 2002 AAAI Spring Symposium Series. AI Mag., vol. 23, no. 4, pp. 101-106, 2002
Fault Detection and Identification in a Mobile Robot using Multiple Model Estimation and Neural Network.
Goel, Puneet; Dedeoglu, Güksel; Roumeliotis, Stergios I.; Sukhatme, Gaurav S.: Fault Detection and Identification in a Mobile Robot using Multiple Model Estimation and Neural Network. Proceedings of the 2000 IEEE International Conference on Robotics and Automation, ICRA 2000, April 24-28, 2000, San Francisco, CA, USA, pp. 2302-2309, IEEE, 2000
On the Development of EMG Control for a Prosthesis Using a Robotic Hand.
Iberall, Thea; Sukhatme, Gaurav S.; Beattie, Denise; Bekey, George A.: On the Development of EMG Control for a Prosthesis Using a Robotic Hand. Proceedings of the 1994 International Conference on Robotics and Automation, San Diego, CA, USA, May 1994, pp. 1753-1758, IEEE Computer Society, 1994
Video
Square grid trajectory flight on a Crazyflie2.1. with a 1 hidden layer 4 neuron neural controller
Square grid trajectory flight on a Crazyflie2.1. with a 1 hidden layer 4 neuron neural controller
Tags: Deep reinforcement learning, Meta learning
Tags: Deep reinforcement learning, Meta learning
A tiny (single hidden layer 4 neuron) neural network, estimated by HyperPPO, driving a quadcopter to track a square grid trajectory.
Bezier Curve trajectory flight on a Crazyflie2.1. with a 1 hidden layer 4 neuron neural controller
Bezier Curve trajectory flight on a Crazyflie2.1. with a 1 hidden layer 4 neuron neural controller
Tags: Deep reinforcement learning, Meta learning
Tags: Deep reinforcement learning, Meta learning
A tiny (single hidden layer 4 neuron) neural network, estimated by HyperPPO, driving a quadcopter to track a 3D random Bezier curve.
Efficiently Learning Small Policies for Locomotion and Manipulation
Efficiently Learning Small Policies for Locomotion and Manipulation
Tags: Deep reinforcement learning, Meta learning
Tags: Deep reinforcement learning, Meta learning
Leveraging graph hyper networks to learn graph hyper policies trained with off-policy reinforcement learning.
Efficient Multi-Task Learning via Iterated Single-Task Transfer
Efficient Multi-Task Learning via Iterated Single-Task Transfer
Tags: Deep reinforcement learning, Meta learning
Tags: Deep reinforcement learning, Meta learning
Multi-task Reinforcement Learning (MTRL) for robots to acquire new skills.
Adaptive Sampling using POMDPs with Domain-Specific Considerations – ICRA 2021
Adaptive Sampling using POMDPs with Domain-Specific Considerations – ICRA 2021
Tags: Informative path planning, Meta learning, Motion planning
Tags: Informative path planning, Meta learning, Motion planning
We investigate improving Monte Carlo Tree Search based solvers for Partially Observable Markov Decision Processes (POMDPs), when applied to adaptive sampling problems. We propose improvements in rollout allocation, the action exploration algorithm, and plan commitment. The first allocates a different number of rollouts depending on how many actions the agent has taken in an episode. We find that rollouts are more valuable after some initial information is gained about the environment. Thus, a linear increase in the number of rollouts, i.e. allocating a fixed number at each step, is not appropriate for adaptive sampling tasks. The second alters which actions the agent chooses to explore when building the planning tree. We find that by using knowledge of the number of rollouts allocated, the agent can more effectively choose actions to explore. The third improvement is in determining how many actions the agent should take from one plan. Typically, an agent will plan to take the first action from the planning tree and then call the planner again from the new state. Using statistical techniques, we show that it is possible to greatly reduce the number of rollouts by increasing the number of actions taken from a single planning tree without affecting the agent’s final reward. Finally, we demonstrate experimentally, on simulated and real aquatic data from an underwater robot, that these improvements can be combined, leading to better adaptive sampling. Accepted at ICRA 2021 The first two authors had an equal contribution
Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors
Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors
Tags: Deep reinforcement learning, Learning from demonstrations, Meta learning
Tags: Deep reinforcement learning, Learning from demonstrations, Meta learning
Quadrotor stabilizing controllers often require careful, model-specific tuning for safe operation. We use reinforcement learning to train policies in a simulation that transfer remarkably well to multiple different physical quadrotors. Our policies are low-level, i.e., we map the rotorcrafts’ state directly to the motor outputs. The trained control policies are very robust to external disturbances and can withstand harsh initial conditions such as throws. To the best of our knowledge, this is the first work that demonstrates that a simple neural network can learn a robust stabilizing low-level quadrotor controller without the use of a stabilizing PD controller; as well as the first work that analyses the transfer capability of a single policy to multiple quadrotors.
