arXiv subjects

Katia Sycara

Publications and source records attributed to Katia Sycara.

At least 91 records · Page 5Linked to original sources

arXiv2021

Emergent Discrete Communication in Semantic Spaces

Neural agents trained in reinforcement learning settings can learn to communicate among themselves via discrete tokens, accomplishing as a team what agents would be unable to do alone. However, the current standard of using one-hot vectors as discrete communication tokens prevents agents from acquiring more desirable aspects of communication such as zero-shot understanding. Inspired by word embedding techniques from natural language processing, we propose neural agent architectures that enables them to communicate via discrete tokens derived from a learned, continuous space. We show in a decision theoretic framework that our technique optimizes communication over a wide range of scenarios, whereas one-hot tokens are only optimal under restrictive assumptions. In self-play experiments, we validate that our trained agents learn to cluster tokens in semantically-meaningful ways, allowing them communicate in noisy environments where other techniques fail. Lastly, we demonstrate both that agents using our method can effectively respond to novel human communication and that humans can understand unlabeled emergent agent communication, outperforming the use of one-hot communication.

cs.LG↗

arXiv2021

Hiding Leader's Identity in Leader-Follower Navigation through Multi-Agent Reinforcement Learning

Leader-follower navigation is a popular class of multi-robot algorithms where a leader robot leads the follower robots in a team. The leader has specialized capabilities or mission-critical information (e.g. goal location) that the followers lack, and this makes the leader crucial for the mission's success. However, this also makes the leader a vulnerability - an external adversary who wishes to sabotage the robot team's mission can simply harm the leader and the whole robot team's mission would be compromised. Since robot motion generated by traditional leader-follower navigation algorithms can reveal the identity of the leader, we propose a defense mechanism of hiding the leader's identity by ensuring the leader moves in a way that behaviorally camouflages it with the followers, making it difficult for an adversary to identify the leader. To achieve this, we combine Multi-Agent Reinforcement Learning, Graph Neural Networks and adversarial training. Our approach enables the multi-robot team to optimize the primary task performance with leader motion similar to follower motion, behaviorally camouflaging it with the followers. Our algorithm outperforms existing work that tries to hide the leader's identity in a multi-robot team by tuning traditional leader-follower control parameters with Classical Genetic Algorithms. We also evaluated human performance in inferring the leader's identity and found that humans had lower accuracy when the robot team used our proposed navigation algorithm.

cs.RO↗

arXiv2021

Learning Embeddings that Capture Spatial Semantics for Indoor Navigation

Incorporating domain-specific priors in search and navigation tasks has shown promising results in improving generalization and sample complexity over end-to-end trained policies. In this work, we study how object embeddings that capture spatial semantic priors can guide search and navigation tasks in a structured environment. We know that humans can search for an object like a book, or a plate in an unseen house, based on the spatial semantics of bigger objects detected. For example, a book is likely to be on a bookshelf or a table, whereas a plate is likely to be in a cupboard or dishwasher. We propose a method to incorporate such spatial semantic awareness in robots by leveraging pre-trained language models and multi-relational knowledge bases as object embeddings. We demonstrate using these object embeddings to search a query object in an unseen indoor environment. We measure the performance of these embeddings in an indoor simulator (AI2Thor). We further evaluate different pre-trained embedding onSuccess Rate(SR) and success weighted by Path Length(SPL).

cs.RO↗

arXiv2021

Deep Interpretable Models of Theory of Mind

When developing AI systems that interact with humans, it is essential to design both a system that can understand humans, and a system that humans can understand. Most deep network based agent-modeling approaches are 1) not interpretable and 2) only model external behavior, ignoring internal mental states, which potentially limits their capability for assistance, interventions, discovering false beliefs, etc. To this end, we develop an interpretable modular neural framework for modeling the intentions of other observed entities. We demonstrate the efficacy of our approach with experiments on data from human participants on a search and rescue task in Minecraft, and show that incorporating interpretability can significantly increase predictive performance under the right conditions.

cs.LG↗

arXiv2021

Online Connectivity-aware Dynamic Deployment for Heterogeneous Multi-Robot Systems

In this paper, we consider the dynamic multi-robot distribution problem where a heterogeneous group of networked robots is tasked to spread out and simultaneously move towards multiple moving task areas while maintaining connectivity. The heterogeneity of the system is characterized by various categories of units and each robot carries different numbers of units per category representing heterogeneous capabilities. Every task area with different importance demands a total number of units contributed by all of the robots within its area. Moreover, we assume the importance and the total number of units requested from each task area is initially unknown. The robots need first to explore, i.e., reach those areas, and then be allocated to the tasks so to fulfill the requirements. The multi-robot distribution problem is formulated as designing controllers to distribute the robots that maximize the overall task fulfillment while minimizing the traveling costs in presence of connectivity constraints. We propose a novel connectivity-aware multi-robot redistribution approach that accounts for dynamic task allocation and connectivity maintenance for a heterogeneous robot team. Such an approach could generate sub-optimal robot controllers so that the amount of total unfulfilled requirements of the tasks weighted by their importance is minimized and robots stay connected at all times. Simulation and numerical results are provided to demonstrate the effectiveness of the proposed approaches.

cs.RO↗

arXiv2021

Adaptive Agent Architecture for Real-time Human-Agent Teaming

Teamwork is a set of interrelated reasoning, actions and behaviors of team members that facilitate common objectives. Teamwork theory and experiments have resulted in a set of states and processes for team effectiveness in both human-human and agent-agent teams. However, human-agent teaming is less well studied because it is so new and involves asymmetry in policy and intent not present in human teams. To optimize team performance in human-agent teaming, it is critical that agents infer human intent and adapt their polices for smooth coordination. Most literature in human-agent teaming builds agents referencing a learned human model. Though these agents are guaranteed to perform well with the learned model, they lay heavy assumptions on human policy such as optimality and consistency, which is unlikely in many real-world scenarios. In this paper, we propose a novel adaptive agent architecture in human-model-free setting on a two-player cooperative game, namely Team Space Fortress (TSF). Previous human-human team research have shown complementary policies in TSF game and diversity in human players' skill, which encourages us to relax the assumptions on human policy. Therefore, we discard learning human models from human data, and instead use an adaptation strategy on a pre-trained library of exemplar policies composed of RL algorithms or rule-based methods with minimal assumptions of human behavior. The adaptation strategy relies on a novel similarity metric to infer human policy and then selects the most complementary policy in our library to maximize the team performance. The adaptive agent architecture can be deployed in real-time and generalize to any off-the-shelf static agents. We conducted human-agent experiments to evaluate the proposed adaptive agent framework, and demonstrated the suboptimality, diversity, and adaptability of human policies in human-agent teams.

cs.RO↗

arXiv2020

Explanation of Reinforcement Learning Model in Dynamic Multi-Agent System

Recently, there has been increasing interest in transparency and interpretability in Deep Reinforcement Learning (DRL) systems. Verbal explanations, as the most natural way of communication in our daily life, deserve more attention, since they allow users to gain a better understanding of the system which ultimately could lead to a high level of trust and smooth collaboration. This paper reports a novel work in generating verbal explanations for DRL behaviors agent. A rule-based model is designed to construct explanations using a series of rules which are predefined with prior knowledge. A learning model is then proposed to expand the implicit logic of generating verbal explanation to general situations by employing rule-based explanations as training data. The learning model is shown to have better flexibility and generalizability than the static rule-based model. The performance of both models is evaluated quantitatively through objective metrics. The results show that verbal explanation generated by both models improve subjective satisfaction of users towards the interpretability of DRL systems. Additionally, seven variants of the learning model are designed to illustrate the contribution of input channels, attention mechanism, and proposed encoder in improving the quality of verbal explanation.

cs.AI↗

arXiv2020

Predicting Human Strategies in Simulated Search and Rescue Task

In a search and rescue scenario, rescuers may have different knowledge of the environment and strategies for exploration. Understanding what is inside a rescuer's mind will enable an observer agent to proactively assist them with critical information that can help them perform their task efficiently. To this end, we propose to build models of the rescuers based on their trajectory observations to predict their strategies. In our efforts to model the rescuer's mind, we begin with a simple simulated search and rescue task in Minecraft with human participants. We formulate neural sequence models to predict the triage strategy and the next location of the rescuer. As the neural networks are data-driven, we design a diverse set of artificial "faux human" agents for training, to test them with limited human rescuer trajectory data. To evaluate the agents, we compare it to an evidence accumulation method that explicitly incorporates all available background knowledge and provides an intended upper bound for the expected performance. Further, we perform experiments where the observer/predictor is human. We show results in terms of prediction accuracy of our computational approaches as compared with that of human observers.

cs.LG↗

arXiv2020

Augmenting GAIL with BC for sample efficient imitation learning

Imitation learning is the problem of recovering an expert policy without access to a reward signal. Behavior cloning and GAIL are two widely used methods for performing imitation learning. Behavior cloning converges in a few iterations but doesn't achieve peak performance due to its inherent iid assumption about the state-action distribution. GAIL addresses the issue by accounting for the temporal dependencies when performing a state distribution matching between the agent and the expert. Although GAIL is sample efficient in the number of expert trajectories required, it is still not very sample efficient in terms of the environment interactions needed for convergence of the policy. Given the complementary benefits of both methods, we present a simple and elegant method to combine both methods to enable stable and sample efficient learning. Our algorithm is very simple to implement and integrates with different policy gradient algorithms. We demonstrate the effectiveness of the algorithm in low dimensional control tasks, gridworlds and in high dimensional image-based tasks.

cs.LG↗

arXiv2020

Feasible Region-based Identification Using Duality (Extended Version)

We consider the problem of estimating bounds on parameters representing tasks being performed by individual robots in a multirobot system. In our previous work, we derived necessary conditions based on persistency of excitation analysis for the exact identification of these parameters. We concluded that depending on the robot's task, the dynamics of individual robots may fail to satisfy these conditions, thereby preventing exact inference. As an extension to that work, this paper focuses on estimating bounds on task parameters when such conditions are not satisfied. Each robot in the team uses optimization-based controllers for mediating between task satisfaction and collision avoidance. We use KKT conditions of this optimization and SVD of active collision avoidance constraints to derive explicit relations between Lagrange multipliers, robot dynamics, and task parameters. Using these relations, we are able to derive bounds on each robot's task parameters. Through numerical simulations, we show how our proposed region based identification approach generates feasible regions for parameters when a conventional estimator such as a UKF fails. Additionally, empirical evidence shows that this approach generates contracting sets which converge to the true parameters much faster than the rate at which a UKF based estimate converges. Videos of these results are available at https://bit.ly/2JDMgeJ

cs.RO↗

arXiv2020

Learning Complex Multi-Agent Policies in Presence of an Adversary

In recent years, there has been some outstanding work on applying deep reinforcement learning to multi-agent settings. Often in such multi-agent scenarios, adversaries can be present. We address the requirements of such a setting by implementing a graph-based multi-agent deep reinforcement learning algorithm. In this work, we consider the scenario of multi-agent deception in which multiple agents need to learn to cooperate and communicate in order to deceive an adversary. We have employed a two-stage learning process to get the cooperating agents to learn such deceptive behaviors. Our experiments show that our approach allows us to employ curriculum learning to increase the number of cooperating agents in the environment and enables a team of agents to learn complex behaviors to successfully deceive an adversary. Keywords: Multi-agent system, Graph neural network, Reinforcement learning

cs.MA↗

arXiv2020

Parameter Identification for Multirobot Systems Using Optimization Based Controllers (Extended Version)

This paper considers the problem of parameter identification for a multirobot system. We wish to understand when is it feasible for an adversarial observer to reverse-engineer the parameters of tasks being performed by a team of robots by simply observing their positions. We address this question by using the concept of persistency of excitation from system identification. Each robot in the team uses optimization-based controllers for mediating between task satisfaction and collision avoidance. These controllers exhibit an implicit dependence on the task's parameters which poses a hurdle for deriving necessary conditions for parameter identification, since such conditions usually require an explicit relation. We address this bottleneck by using duality theory and SVD of active collision avoidance constraints and derive an explicit relation between each robot's task parameters and its control inputs. This allows us to derive the main necessary conditions for successful identification which agree with our intuition. We demonstrate the importance of these conditions through numerical simulations by using (a) an adaptive observer and (b) an unscented Kalman filter for goal estimation in various geometric settings. These simulations show that under circumstances where parameter inference is supposed to be infeasible per our conditions, both these estimators fail and likewise when it is feasible, both converge to the true parameters. Videos of these results are available at https://bit.ly/3kQYj5J.

math.OC↗

arXiv2020

Addressing reward bias in Adversarial Imitation Learning with neutral reward functions

Generative Adversarial Imitation Learning suffers from the fundamental problem of reward bias stemming from the choice of reward functions used in the algorithm. Different types of biases also affect different types of environments - which are broadly divided into survival and task-based environments. We provide a theoretical sketch of why existing reward functions would fail in imitation learning scenarios in task based environments with multiple terminal states. We also propose a new reward function for GAIL which outperforms existing GAIL methods on task based environments with single and multiple terminal states and effectively overcomes both survival and termination bias.

cs.LG↗

arXiv2020

Deadlock Analysis and Resolution in Multi-Robot Systems (Extended Version)

Collision avoidance for multirobot systems is a well studied problem. Recently, control barrier functions (CBFs) have been proposed for synthesizing controllers guarantee collision avoidance and goal stabilization for multiple robots. However, it has been noted reactive control synthesis methods (such as CBFs) are prone to deadlock, an equilibrium of system dynamics causes robots to come to a standstill before reaching their goals. In this paper, we formally derive characteristics of deadlock in a multirobot system uses CBFs. We propose a novel approach to analyze deadlocks resulting from optimization based controllers (CBFs) by borrowing tools from duality theory and graph enumeration. Our key insight is system deadlock is characterized by a force-equilibrium on robots and we show how complexity of deadlock analysis increases approximately exponentially with the number of robots. This analysis allows us to interpret deadlock as a subset of the state space, and we prove this set is non-empty, bounded and located on the boundary of the safety set. Finally, we use these properties to develop a provably correct decentralized algorithm for deadlock resolution which ensures robots converge to their goals while avoiding collisions. We show simulation results of the resolution algorithm for two and three robots and experimentally validate this algorithm on Khepera-IV robots.

cs.RO↗

arXiv2020

Natural Emergence of Heterogeneous Strategies in Artificially Intelligent Competitive Teams

Multi agent strategies in mixed cooperative-competitive environments can be hard to craft by hand because each agent needs to coordinate with its teammates while competing with its opponents. Learning based algorithms are appealing but many scenarios require heterogeneous agent behavior for the team's success and this increases the complexity of the learning algorithm. In this work, we develop a competitive multi agent environment called FortAttack in which two teams compete against each other. We corroborate that modeling agents with Graph Neural Networks and training them with Reinforcement Learning leads to the evolution of increasingly complex strategies for each team. We observe a natural emergence of heterogeneous behavior amongst homogeneous agents when such behavior can lead to the team's success. Such heterogeneous behavior from homogeneous agents is appealing because any agent can replace the role of another agent at test time. Finally, we propose ensemble training, in which we utilize the evolved opponent strategies to train a single policy for friendly agents.

cs.AI↗

arXiv2020

MA 3 : Model Agnostic Adversarial Augmentation for Few Shot learning

Despite the recent developments in vision-related problems using deep neural networks, there still remains a wide scope in the improvement of generalizing these models to unseen examples. In this paper, we explore the domain of few-shot learning with a novel augmentation technique. In contrast to other generative augmentation techniques, where the distribution over input images are learnt, we propose to learn the probability distribution over the image transformation parameters which are easier and quicker to learn. Our technique is fully differentiable which enables its extension to versatile data-sets and base models. We evaluate our proposed method on multiple base-networks and 2 data-sets to establish the robustness and efficiency of this method. We obtain an improvement of nearly 4% by adding our augmentation module without making any change in network architectures. We also make the code readily available for usage by the community.

cs.CV↗

arXiv2020

Behavior Mixing with Minimum Global and Subgroup Connectivity Maintenance for Large-Scale Multi-Robot Systems

In many cases the multi-robot systems are desired to execute simultaneously multiple behaviors with different controllers, and sequences of behaviors in real time, which we call \textit{behavior mixing}. Behavior mixing is accomplished when different subgroups of the overall robot team change their controllers to collectively achieve given tasks while maintaining connectivity within and across subgroups in one connected communication graph. In this paper, we present a provably minimum connectivity maintenance framework to ensure the subgroups and overall robot team stay connected at all times while providing the highest freedom for behavior mixing. In particular, we propose a real-time distributed Minimum Connectivity Constraint Spanning Tree (MCCST) algorithm to select the minimum inter-robot connectivity constraints preserving subgroup and global connectivity that are \textit{least likely to be violated} by the original controllers. With the employed safety and connectivity barrier certificates for the activated connectivity constraints and collision avoidance, the behavior mixing controllers are thus minimally modified from the original controllers. We demonstrate the effectiveness and scalability of our approach via simulations of up to 100 robots with multiple behaviors.

cs.RO↗

arXiv2019

Simultaneous Identification of Tweet Purpose and Position

Tweet classification has attracted considerable attention recently. Most of the existing work on tweet classification focuses on topic classification, which classifies tweets into several predefined categories, and sentiment classification, which classifies tweets into positive, negative and neutral. Since tweets are different from conventional text in that they generally are of limited length and contain informal, irregular or new words, so it is difficult to determine user intention to publish a tweet and user attitude towards certain topic. In this paper, we aim to simultaneously classify tweet purpose, i.e., the intention for user to publish a tweet, and position, i.e., supporting, opposing or being neutral to a given topic. By transforming this problem to a multi-label classification problem, a multi-label classification method with post-processing is proposed. Experiments on real-world data sets demonstrate the effectiveness of this method and the results outperform the individual classification methods.

cs.CL↗