Mohit Sharma

Homanga Bharadhwaj, Jay Vakil, Mohit Sharma et al.

The grand aim of having a single robot that can manipulate arbitrary objects in diverse settings is at odds with the paucity of robotics datasets. Acquiring and growing such datasets is strenuous due to manual efforts, operational costs, and safety challenges. A path toward such an universal agent would require a structured framework capable of wide generalization but trained within a reasonable data budget. In this paper, we develop an efficient system (RoboAgent) for training universal agents capable of multi-task manipulation skills using (a) semantic augmentations that can rapidly multiply existing datasets and (b) action representations that can extract performant policies with small yet diverse multi-modal datasets without overfitting. In addition, reliable task conditioning and an expressive policy architecture enable our agent to exhibit a diverse repertoire of skills in novel situations specified using language commands. Using merely 7500 demonstrations, we are able to train a single agent capable of 12 unique skills, and demonstrate its generalization over 38 tasks spread across common daily activities in diverse kitchen scenes. On average, RoboAgent outperforms prior methods by over 40% in unseen situations while being more sample efficient and being amenable to capability improvements and extensions through fine-tuning. Videos at https://robopen.github.io/

Zoey Chen, Zhao Mandi, Homanga Bharadhwaj et al.

Generalization to unseen real-world scenarios for robot manipulation requires exposure to diverse datasets during training. However, collecting large real-world datasets is intractable due to high operational costs. For robot learning to generalize despite these challenges, it is essential to leverage sources of data or priors beyond the robot's direct experience. In this work, we posit that image-text generative models, which are pre-trained on large corpora of web-scraped data, can serve as such a data source. These generative models encompass a broad range of real-world scenarios beyond a robot's direct experience and can synthesize novel synthetic experiences that expose robotic agents to additional world priors aiding real-world generalization at no extra cost. In particular, our approach leverages pre-trained generative models as an effective tool for data augmentation. We propose a generative augmentation framework for semantically controllable augmentations and rapidly multiplying robot datasets while inducing rich variations that enable real-world generalization. Based on diverse augmentations of robot data, we show how scalable robot manipulation policies can be trained and deployed both in simulation and in unseen real-world environments such as kitchens and table-tops. By demonstrating the effectiveness of image-text generative models in diverse real-world robotic applications, our generative augmentation framework provides a scalable and efficient path for boosting generalization in robot learning at no extra human cost.

Anirudha Majumdar, Mohit Sharma, Dmitry Kalashnikov et al.

Visuomotor policies trained via imitation learning are capable of performing challenging manipulation tasks, but are often extremely brittle to lighting, visual distractors, and object locations. These vulnerabilities can depend unpredictably on the specifics of training, and are challenging to expose without time-consuming and expensive hardware evaluations. We propose the problem of predictive red teaming: discovering vulnerabilities of a policy with respect to environmental factors, and predicting the corresponding performance degradation without hardware evaluations in off-nominal scenarios. In order to achieve this, we develop RoboART: an automated red teaming (ART) pipeline that (1) modifies nominal observations using generative image editing to vary different environmental factors, and (2) predicts performance under each variation using a policy-specific anomaly detector executed on edited observations. Experiments across 500+ hardware trials in twelve off-nominal conditions for visuomotor diffusion policies demonstrate that RoboART predicts performance degradation with high accuracy (less than 0.19 average difference between predicted and real success rates). We also demonstrate how predictive red teaming enables targeted data collection: fine-tuning with data collected under conditions predicted to be adverse boosts baseline performance by 2-7x.

Connor Schenck, Isaac Reid, Mithun George Jacob et al.

We introduce STRING: Separable Translationally Invariant Position Encodings. STRING extends Rotary Position Encodings, a recently proposed and widely used algorithm in large language models, via a unifying theoretical framework. Importantly, STRING still provides exact translation invariance, including token coordinates of arbitrary dimensionality, whilst maintaining a low computational footprint. These properties are especially important in robotics, where efficient 3D token representation is key. We integrate STRING into Vision Transformers with RGB(-D) inputs (color plus optional depth), showing substantial gains, e.g. in open-vocabulary object detection and for robotics controllers. We complement our experiments with a rigorous mathematical analysis, proving the universality of our methods.

Mohit Sharma, Adam Fishman, Vikash Kumar et al.

Robots in the real world need to perceive and move to goals in complex environments without collisions. Avoiding collisions is especially difficult when relying on sensor perception and when goals are among clutter. Diffusion policies and other generative models have shown strong performance in solving local planning problems, but often struggle at avoiding all of the subtle constraint violations that characterize truly challenging global motion planning problems. In this work, we propose an approach for learning global motion planning using diffusion policies, allowing the robot to generate full trajectories through complex scenes and reasoning about multiple obstacles along the path. Our approach uses cascaded hierarchical models which unify global prediction and local refinement together with online plan repair to ensure the trajectories are collision free. Our method outperforms (by ~5%) a wide variety of baselines on challenging tasks in multiple domains including navigation and manipulation.

Saumya Saxena, Mohit Sharma, Oliver Kroemer

Leveraging sensing modalities across diverse spatial and temporal resolutions can improve performance of robotic manipulation tasks. Multi-spatial resolution sensing provides hierarchical information captured at different spatial scales and enables both coarse and precise motions. Simultaneously multi-temporal resolution sensing enables the agent to exhibit high reactivity and real-time control. In this work, we propose a framework, MResT (Multi-Resolution Transformer), for learning generalizable language-conditioned multi-task policies that utilize sensing at different spatial and temporal resolutions using networks of varying capacities to effectively perform real time control of precise and reactive tasks. We leverage off-the-shelf pretrained vision-language models to operate on low-frequency global features along with small non-pretrained models to adapt to high frequency local feedback. Through extensive experiments in 3 domains (coarse, precise and dynamic manipulation tasks), we show that our approach significantly improves (2X on average) over recent multi-task baselines. Further, our approach generalizes well to visual and geometric variations in target objects and to varying interaction forces.

Mohit Sharma, Oliver Kroemer

To perform manipulation tasks in the real world, robots need to operate on objects with various shapes, sizes and without access to geometric models. It is often unfeasible to train monolithic neural network policies across such large variance in object properties. Towards this generalization challenge, we propose to learn modular task policies which compose object-centric task-axes controllers. These task-axes controllers are parameterized by properties associated with underlying objects in the scene. We infer these controller parameters directly from visual input using multi-view dense correspondence learning. Our overall approach provides a simple, modular and yet powerful framework for learning manipulation tasks. We empirically evaluate our approach on multiple different manipulation tasks and show its ability to generalize to large variance in object size, shape and geometry.

Navyata Sanghvi, Shinnosuke Usami, Mohit Sharma et al.

Various methods for solving the inverse reinforcement learning (IRL) problem have been developed independently in machine learning and economics. In particular, the method of Maximum Causal Entropy IRL is based on the perspective of entropy maximization, while related advances in the field of economics instead assume the existence of unobserved action shocks to explain expert behavior (Nested Fixed Point Algorithm, Conditional Choice Probability method, Nested Pseudo-Likelihood Algorithm). In this work, we make previously unknown connections between these related methods from both fields. We achieve this by showing that they all belong to a class of optimization problems, characterized by a common form of the objective, the associated policy and the objective gradient. We demonstrate key computational and algorithmic differences which arise between the methods due to an approximation of the optimal soft value function, and describe how this leads to more efficient algorithms. Using insights which emerge from our study of this class of optimization problems, we identify various problem scenarios and investigate each method's suitability for these problems.

Saurav Manchanda, Mohit Sharma, George Karypis

Slot-filling refers to the task of annotating individual terms in a query with the corresponding intended product characteristics (product type, brand, gender, size, color, etc.). These characteristics can then be used by a search engine to return results that better match the query's product intent. Traditional methods for slot-filling require the availability of training data with ground truth slot-annotation information. However, generating such labeled data, especially in e-commerce is expensive and time-consuming because the number of slots increases as new products are added. In this paper, we present distant-supervised probabilistic generative models, that require no manual annotation. The proposed approaches leverage the readily available historical query logs and the purchases that these queries led to, and also exploit co-occurrence information among the slots in order to identify intended product characteristics. We evaluate our approaches by considering how they affect retrieval performance, as well as how well they classify the slots. In terms of retrieval, our approaches achieve better ranking performance (up to 156%) over Okapi BM25. Moreover, our approach that leverages co-occurrence information leads to better performance than the one that does not on both the retrieval and slot classification tasks.

Mohit Sharma, Oliver Kroemer

To determine if a skill can be executed in any given environment, a robot needs to learn the preconditions for the skill. As robots begin to operate in dynamic and unstructured environments, precondition models will need to generalize to variable number of objects with different shapes and sizes. In this work, we focus on learning precondition models for manipulation skills in unconstrained environments. Our work is motivated by the intuition that many complex manipulation tasks, with multiple objects, can be simplified by focusing on less complex pairwise object relations. We propose an object-relation model that learns continuous representations for these pairwise object relations. Our object-relation model is trained completely in simulation, and once learned, is used by a separate precondition model to predict skill preconditions for real world tasks. We evaluate our precondition model on $3$ different manipulation tasks: sweeping, cutting, and unstacking. We show that our approach leads to significant improvements in predicting preconditions for all 3 tasks, across objects of different shapes and sizes.

Kevin Zhang, Mohit Sharma, Manuela Veloso et al.

Cutting is a common form of manipulation when working with divisible objects such as food, rope, or clay. Cooking in particular relies heavily on cutting to divide food items into desired shapes. However, cutting food is a challenging task due to the wide range of material properties exhibited by food items. Due to this variability, the same cutting motions cannot be used for all food items. Sensations from contact events, e.g., when placing the knife on the food item, will also vary depending on the material properties, and the robot will need to adapt accordingly. In this paper, we propose using vibrations and force-torque feedback from the interactions to adapt the slicing motions and monitor for contact events. The robot learns neural networks for performing each of these tasks and generalizing across different material properties. By adapting and monitoring the skill executions, the robot is able to reliably cut through more than 20 different types of food items and even detect whether certain food items are fresh or old.

Saurav Manchanda, Mohit Sharma, George Karypis

In e-commerce, a user tends to search for the desired product by issuing a query to the search engine and examining the retrieved results. If the search engine was successful in correctly understanding the user's query, it will return results that correspond to the products whose attributes match the terms in the query that are representative of the query's product intent. However, the search engine may fail to retrieve results that satisfy the query's product intent and thus degrading user experience due to different issues in query processing: (i) when multiple terms are present in a query it may fail to determine the relevant terms that are representative of the query's product intent, and (ii) it may suffer from vocabulary gap between the terms in the query and the product's description, i.e., terms used in the query are semantically similar but different from the terms in the product description. Hence, identifying the terms that describe the query's product intent and predicting additional terms that describe the query's product intent better than the existing query terms to the search engine is an essential task in e-commerce search. In this paper, we leverage the historical query reformulation logs of a major e-commerce retailer to develop distant-supervised approaches to solve both these problems. Our approaches exploit the fact that the significance of a term is dependent upon the context (other terms in the neighborhood) in which it is used in order to learn the importance of the term towards the query's product intent. We show that identifying and emphasizing the terms that define the query's product intent leads to a 3% improvement in ranking. Moreover, for the tasks of identifying the important terms in a query and for predicting the additional terms that represent product intent, experiments illustrate that our approaches outperform the non-contextual baselines.

Nirbhay Modhe, Prithvijit Chattopadhyay, Mohit Sharma et al.

We propose a novel framework to identify sub-goals useful for exploration in sequential decision making tasks under partial observability. We utilize the variational intrinsic control framework (Gregor et.al., 2016) which maximizes empowerment -- the ability to reliably reach a diverse set of states and show how to identify sub-goals as states with high necessary option information through an information theoretic regularizer. Despite being discovered without explicit goal supervision, our sub-goals provide better exploration and sample complexity on challenging grid-world navigation tasks compared to supervised counterparts in prior work.

Siddharth Gururani, Mohit Sharma, Alexander Lerch

While the automatic recognition of musical instruments has seen significant progress, the task is still considered hard for music featuring multiple instruments as opposed to single instrument recordings. Datasets for polyphonic instrument recognition can be categorized into roughly two categories. Some, such as MedleyDB, have strong per-frame instrument activity annotations but are usually small in size. Other, larger datasets such as OpenMIC only have weak labels, i.e., instrument presence or absence is annotated only for long snippets of a song. We explore an attention mechanism for handling weakly labeled data for multi-label instrument recognition. Attention has been found to perform well for other tasks with weakly labeled data. We compare the proposed attention model to multiple models which include a baseline binary relevance random forest, recurrent neural network, and fully connected neural networks. Our results show that incorporating attention leads to an overall improvement in classification accuracy metrics across all 20 instruments in the OpenMIC dataset. We find that attention enables models to focus on (or `attend to') specific time segments in the audio relevant to each instrument label leading to interpretable results.

Mohit Sharma, F. Maxwell Harper, George Karypis

Most of the existing recommender systems use the ratings provided by users on individual items. An additional source of preference information is to use the ratings that users provide on sets of items. The advantages of using preferences on sets are two-fold. First, a rating provided on a set conveys some preference information about each of the set's items, which allows us to acquire a user's preferences for more items that the number of ratings that the user provided. Second, due to privacy concerns, users may not be willing to reveal their preferences on individual items explicitly but may be willing to provide a single rating to a set of items, since it provides some level of information hiding. This paper investigates two questions related to using set-level ratings in recommender systems. First, how users' item-level ratings relate to their set-level ratings. Second, how collaborative filtering-based models for item-level rating prediction can take advantage of such set-level ratings. We have collected set-level ratings from active users of Movielens on sets of movies that they have rated in the past. Our analysis of these ratings shows that though the majority of the users provide the average of the ratings on a set's constituent items as the rating on the set, there exists a significant number of users that tend to consistently either under- or over-rate the sets. We have developed collaborative filtering-based methods to explicitly model these user behaviors that can be used to recommend items to users. Experiments on real data and on synthetic data that resembles the under- or over-rating behavior in the real data, demonstrate that these models can recover the overall characteristics of the underlying data and predict the user's ratings on individual items.

Mohit Sharma, Kevin Zhang, Oliver Kroemer

In this work, we present an interaction-based approach to learn semantically rich representations for the task of slicing vegetables. Unlike previous approaches, we focus on object-centric representations and use auxiliary tasks to learn rich representations using a two-step process. First, we use simple auxiliary tasks, such as predicting the thickness of a cut slice, to learn an embedding space which captures object properties that are important for the task of slicing vegetables. In the second step, we use these learned latent embeddings to learn a forward model. Learning a forward model affords us to plan online in the latent embedding space and forces our model to improve its representations while performing the slicing task. To show the efficacy of our approach we perform experiments on two different vegetables: cucumbers and tomatoes. Our experimental evaluation shows that our method is able to capture important semantic properties for the slicing task, such as the thickness of the vegetable being cut. We further show that by using our learned forward model, we can plan for the task of vegetable slicing.

Arjun Sharma, Mohit Sharma, Nicholas Rhinehart et al.

The use of imitation learning to learn a single policy for a complex task that has multiple modes or hierarchical structure can be challenging. In fact, previous work has shown that when the modes are known, learning separate policies for each mode or sub-task can greatly improve the performance of imitation learning. In this work, we discover the interaction between sub-tasks from their resulting state-action trajectory sequences using a directed graphical model. We propose a new algorithm based on the generative adversarial imitation learning framework which automatically learns sub-task policies from unsegmented demonstrations. Our approach maximizes the directed information flow in the graphical model between sub-task latent variables and their generated trajectories. We also show how our approach connects with the existing Options framework, which is commonly used to learn hierarchical policies.

Mohit Sharma, Kris M. Kitani, Joachim Groeger

We make an important connection to existing results in econometrics to describe an alternative formulation of inverse reinforcement learning (IRL). In particular, we describe an algorithm using Conditional Choice Probabilities (CCP), which are maximum likelihood estimates of the policy estimated from expert demonstrations, to solve the IRL problem. Using the language of structural econometrics, we re-frame the optimal decision problem and introduce an alternative representation of value functions due to (Hotz and Miller 1993). In addition to presenting the theoretical connections that bridge the IRL literature between Economics and Robotics, the use of CCPs also has the practical benefit of reducing the computational cost of solving the IRL problem. Specifically, under the CCP representation, we show how one can avoid repeated calls to the dynamic programming subroutine typically used in IRL. We show via extensive experimentation on standard IRL benchmarks that CCP-IRL is able to outperform MaxEnt-IRL, with as much as a 5x speedup and without compromising on the quality of the recovered reward function.