Ajay Kumar Tanwani

Ajay Kumar Tanwani

Auto-generated profile, based on publicly available information

Ajay Kumar Tanwani

University of California, Berkeley, CA, USA

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 May 2020 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
ATAjay Kumar Tanwani
Ajay Kumar Tanwani
RARaghav Anand
Raghav Anand
JGJoseph E. Gonzalez
Joseph E. Gonzalez
KGKen Goldberg
Ken Goldberg
Programming robots is complicated due to the lack of ‘plug-and-play’ modules for skill acquisition. Virtualizing deployment of deep learning models can facilitate large-scale use/re-use of off-the-shelf functional behaviors. Deploying deep learning models on robots entails real-time, accurate and reliable inference service under varying query load. This letter introduces a novel Robot-Inference-and-Learning-as-a-Service (RILaaS) platform for low-latency and secure inference serving of deep models that can be deployed on robots. Unique features of RILaaS include: 1) low-latency and reliable serving with gRPC under dynamic loads by distributing queries over multiple servers on Edge and Cloud, 2) SSH based authentication coupled with SSL/TLS based encryption for security and privacy of the data, and 3) front-end REST API for sharing, monitoring and visualizing performance metrics of the available models. We report experiments to evaluate the RILaaS platform under varying loads of batch size, number of robots, and various model placement hosts on Cloud, Edge, and Fog for providing benchmark applications of object recognition and grasp planning as a service. We address the complexity of load balancing with a reinforcement learning algorithm that optimizes simulated profiles of networked robots; outperforming several baselines including round robin, least connections, and least model time with 68.30% and 14.04% decrease in round-trip latency time across models compared to the worst and the next best baseline respectively. Details and updates are available at: https://sites.google.com/view/rilaas

6 August 2021 in The International Journal of Robotics Research
Verified
Journal Article
Open access
PDF file
ATAjay Kumar Tanwani
Ajay Kumar Tanwani
AYAndy Yan
Andy Yan
JLJonathan Lee
Jonathan Lee
SCSylvain Calinon
Sylvain Calinon
KGKen Goldberg
Ken Goldberg
This paper presents a framework to learn the sequential structure in the demonstrations for robot imitation learning. We first present a family of task-parameterized hidden semi-Markov models that extracts invariant segments (also called sub-goals or options) from demonstrated trajectories, and optimally follows the sampled sequence of states from the model with a linear quadratic tracking controller. We then extend the concept to learning invariant segments from visual observations that are sequenced together for robot imitation. We present Motion2Vec that learns a deep embedding space by minimizing a metric learning loss in a Siamese network: images from the same action segment are pulled together while being pushed away from randomly sampled images of other segments, and a time contrastive loss is used to preserve the temporal ordering of the images. The trained embeddings are segmented with a recurrent neural network, and subsequently used for decoding the end-effector pose of the robot. We first show its application to a pick-and-place task with the Baxter robot while avoiding a moving obstacle from four kinesthetic demonstrations only, followed by suturing task imitation from publicly available suturing videos of the JIGSAWS dataset with state-of-the-art 85.5 % segmentation accuracy and 0.94 cm error in position per observation on the test set.

11 December 2018 in The International Journal of Robotics Research
Verified
Journal Article
Closed access
ATAjay Kumar Tanwani
Ajay Kumar Tanwani
SCSylvain Calinon
Sylvain Calinon
Small-variance asymptotics is emerging as a useful technique for inference in large-scale Bayesian non-parametric mixture models. This paper analyzes the online learning of robot manipulation tasks with Bayesian non-parametric mixture models under small-variance asymptotics. The analysis yields a scalable online sequence clustering (SOSC) algorithm that is non-parametric in the number of clusters and the subspace dimension of each cluster. SOSC groups the new datapoint in low-dimensional subspaces by online inference in a non-parametric mixture of probabilistic principal component analyzers (MPPCA) based on a Dirichlet process, and captures the state transition and state duration information online in a hidden semi-Markov model (HSMM) based on a hierarchical Dirichlet process. A task-parameterized formulation of our approach autonomously adapts the model to changing environmental situations during manipulation. We apply the algorithm in a teleoperation setting to recognize the intention of the operator and remotely adjust the movement of the robot using the learned model. The generative model is used to synthesize both time-independent and time-dependent behaviors by relying on the principles of shared and autonomous control. Experiments with the Baxter robot yield parsimonious clusters that adapt online with new demonstrations and assist the operator in performing remote manipulation tasks.

24 January 2021 in The International Journal of Robotics Research
Verified
Journal Article
Closed access
PDF file
JLJonathan N. Lee
Jonathan N. Lee
MLMichael Laskey
Michael Laskey
ATAjay Kumar Tanwani
Ajay Kumar Tanwani
AAAnil Aswani
Anil Aswani
KGKen Goldberg
Ken Goldberg
On-policy imitation learning algorithms such as DAgger evolve a robot control policy by executing it, measuring performance (loss), obtaining corrective feedback from a supervisor, and generating the next policy. As the loss between iterations can vary unpredictably, a fundamental question is under what conditions this process will eventually achieve a converged policy. If one assumes the underlying trajectory distribution is static (stationary), it is possible to prove convergence for DAgger. However, in more realistic models for robotics, the underlying trajectory distribution is dynamic because it is a function of the policy. Recent results show it is possible to prove convergence of DAgger when a regularity condition on the rate of change of the trajectory distributions is satisfied. In this article, we reframe this result using dynamic regret theory from the field of online optimization and show that dynamic regret can be applied to any on-policy algorithm to analyze its convergence and optimality. These results inspire a new algorithm, Adaptive On-Policy Regularization (Aor), that ensures the conditions for convergence. We present simulation results with cart–pole balancing and locomotion benchmarks that suggest Aor can significantly decrease dynamic regret and chattering as the robot learns. To the best of the authors’ knowledge, this is the first application of dynamic regret theory to imitation learning.

8 May 2020
Verified
Book Chapter
Closed access
PDF file
ATAjay Kumar Tanwani
Ajay Kumar Tanwani
JLJonathan Lee
Jonathan Lee
BTBrijen Thananjeyan
Brijen Thananjeyan
MLMichael Laskey
Michael Laskey
SKSanjay Krishnan
Sanjay Krishnan
SCSylvain Calinon
Sylvain Calinon

1 August 2020
Verified
Conference Paper
Open access
KSKate Sanders
Kate Sanders
MDMichael Danielczuk
Michael Danielczuk
JMJeffrey Mahler
Jeffrey Mahler
ATAjay Tanwani
Ajay Tanwani
KGKen Goldberg
Ken Goldberg
A new generation of automated bin picking systems using deep learning is evolving to support increasing demand for e-commerce. To accommodate a wide variety of products, many automated systems include multiple gripper types and/or tool changers. However, for some objects, sequential grasp failures are common: when a computed grasp fails to lift and remove the object, the bin is often left unchanged; as the sensor input is consistent, the system retries the same grasp over and over, resulting in a significant reduction in mean successful picks per hour (MPPH). Based on an empirical study of sequential failures, we characterize a class of “sequential failure objects” (SFOs) – objects prone to sequential failures based on a novel taxonomy. We then propose three non-Markov picking policies that incorporate memory of past failures to modify subsequent actions. Simulation experiments on SFO models and the EGAD dataset [19] suggest that the non-Markov policies significantly outperform the Markov policy in terms of the sequential failure rate and MPPH. In physical experiments on 50 heaps of 12 SFOs the most effective Non-Markov policy increased MPPH over the Dex-Net Markov policy by 107%.