Motion Planning with Competency-Aware Transition Models for Underactuated Adaptive Hands

Underactuated adaptive hands simplify grasping tasks but it is difficult to model their interactions with objects during in-hand manipulation. Learned data-driven models have been recently shown to be efficient in motion planning and control of such hands. Still, the accuracy of the models is limited even with the addition of more data. This becomes important for long horizon predictions, where errors are accumulated along the length of a path. Instead of throwing more data into learning the transition model, this work proposes to rather invest a portion of the training data in a critic model. The critic is trained to estimate the error of the transition model given a state and a sequence of future actions, along with information of past actions. The critic is used to reformulate the cost function of an asymptotically optimal motion planner. Given the critic, the planner directs planned paths to less erroneous regions in the state space. The approach is evaluated against standard motion planning on simulated and real hands. The results show that it outperforms an alternative where all the available data is used for training the transition model without a critic.

[1]  Burr Settles,et al.  Active Learning Literature Survey , 2009 .

[2]  Yarin Gal,et al.  Uncertainty in Deep Learning , 2016 .

[3]  Ross B. Girshick,et al.  Mask R-CNN , 2017, 1703.06870.

[4]  Kostas E. Bekris,et al.  Belief-Space Planning Using Learned Models with Application to Underactuated Hands , 2019, ISRR.

[5]  Aaron M. Dollar,et al.  Learning Modes of Within-Hand Manipulation , 2018, 2018 IEEE International Conference on Robotics and Automation (ICRA).

[6]  Christopher G. Atkeson,et al.  Neural networks and differential dynamic programming for reinforcement learning problems , 2016, 2016 IEEE International Conference on Robotics and Automation (ICRA).

[7]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[8]  Kostas E. Bekris,et al.  Path Planning for Within-Hand Manipulation over Learned Representations of Safe States , 2018, ISER.

[9]  Yarin Gal,et al.  Understanding Measures of Uncertainty for Adversarial Example Detection , 2018, UAI.

[10]  Manuel G. Catalano,et al.  Adaptive synergies: An approach to the design of under-actuated robotic hands , 2012, 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[11]  Kostas E. Bekris,et al.  Model Identification via Physics Engines for Improved Policy Search , 2017, ArXiv.

[12]  Clément Gosselin,et al.  Simulation and design of underactuated mechanical hands , 1998 .

[13]  Aaron M. Dollar,et al.  Stable, open-loop precision manipulation with underactuated hands , 2015, Int. J. Robotics Res..

[14]  Kostas E. Bekris,et al.  Efficient Model Identification for Tensegrity Locomotion , 2018, 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[15]  Aaron M. Dollar,et al.  Vision-based precision manipulation with underactuated hands: Simple and effective solutions for dexterity , 2016, 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[16]  Aaron M. Dollar,et al.  A parallel robots framework to study precision grasping and dexterous manipulation , 2013, 2013 IEEE International Conference on Robotics and Automation.

[17]  Marta Z. Kwiatkowska,et al.  Evaluating Uncertainty Quantification in End-to-End Autonomous Driving Control , 2018, ArXiv.

[18]  Aaron M. Dollar,et al.  Dexterous manipulation with underactuated elastic hands , 2011, 2011 IEEE International Conference on Robotics and Automation.

[19]  Kostas E. Bekris,et al.  Informed Asymptotically Near-Optimal Planning for Field Robots with Dynamics , 2017, FSR.

[20]  Kostas E. Bekris,et al.  The Importance of a Suitable Distance Function in Belief-Space Planning , 2015, ISRR.

[21]  Aaron Dollar,et al.  Yale OpenHand Project: Optimizing Open-Source Hand Designs for Ease of Fabrication and Adoption , 2017, IEEE Robotics & Automation Magazine.

[22]  Dongming Gan,et al.  Modeling and Prototyping of an Underactuated Gripper Exploiting Joint Compliance and Modularity , 2018, IEEE Robotics and Automation Letters.

[23]  Jan Peters,et al.  Active Incremental Learning of Robot Movement Primitives , 2017, CoRL.

[24]  Antonio Bicchi,et al.  On the manipulability ellipsoids of underactuated robotic hands with compliance , 2012, Robotics Auton. Syst..

[25]  Jan Peters,et al.  Learning robot in-hand manipulation with tactile features , 2015, 2015 IEEE-RAS 15th International Conference on Humanoid Robots (Humanoids).

[26]  Alessio Rocchi,et al.  Stable simulation of underactuated compliant hands , 2016, 2016 IEEE International Conference on Robotics and Automation (ICRA).

[27]  Matthew T. Mason,et al.  Data-driven statistical modeling of a cube regrasp , 2016, 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).

[28]  Athanasios S. Polydoros,et al.  Survey of Model-Based Reinforcement Learning: Applications on Robotics , 2017, J. Intell. Robotic Syst..

[29]  Kostas E. Bekris,et al.  Learning a State Transition Model of an Underactuated Adaptive Hand , 2019, IEEE Robotics and Automation Letters.

[30]  Siddhartha S. Srinivasa,et al.  A data-driven statistical framework for post-grasp manipulation , 2014, Int. J. Robotics Res..

[31]  Jochen J. Steil,et al.  Hybrid Analytical and Data-Driven Modeling for Feed-Forward Robot Control † , 2017, Sensors.

[32]  Andreas Krause,et al.  Near-Optimal Sensor Placements in Gaussian Processes: Theory, Efficient Algorithms and Empirical Studies , 2008, J. Mach. Learn. Res..

[33]  Abdeslam Boularias,et al.  Learning to Manipulate Unknown Objects in Clutter by Reinforcement , 2015, AAAI.

[34]  Zoubin Ghahramani,et al.  Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning , 2015, ICML.

[35]  Kostas E. Bekris,et al.  Efficient and Asymptotically Optimal Kinodynamic Motion Planning via Dominance-Informed Regions , 2018, 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS).