arXiv subjects

Sandeep Chinchali

Publications and source records attributed to Sandeep Chinchali.

At least 55 records · Page 3Linked to original sources

arXiv2023

Data Games: A Game-Theoretic Approach to Swarm Robotic Data Collection

Fleets of networked autonomous vehicles (AVs) collect terabytes of sensory data, which is often transmitted to central servers (the ''cloud'') for training machine learning (ML) models. Ideally, these fleets should upload all their data, especially from rare operating contexts, in order to train robust ML models. However, this is infeasible due to prohibitive network bandwidth and data labeling costs. Instead, we propose a cooperative data sampling strategy where geo-distributed AVs collaborate to collect a diverse ML training dataset in the cloud. Since the AVs have a shared objective but minimal information about each other's local data distribution and perception model, we can naturally cast cooperative data collection as an $N$-player mathematical game. We show that our cooperative sampling strategy uses minimal information to converge to a centralized oracle policy with complete information about all AVs. Moreover, we theoretically characterize the performance benefits of our game-theoretic strategy compared to greedy sampling. Finally, we experimentally demonstrate that our method outperforms standard benchmarks by up to $21.9\%$ on 4 perception datasets, including for autonomous driving in adverse weather conditions. Crucially, our experimental results on real-world datasets closely align with our theoretical guarantees.

cs.AI↗

arXiv2023

Learning Sparse Control Tasks from Pixels by Latent Nearest-Neighbor-Guided Explorations

Recent progress in deep reinforcement learning (RL) and computer vision enables artificial agents to solve complex tasks, including locomotion, manipulation and video games from high-dimensional pixel observations. However, domain specific reward functions are often engineered to provide sufficient learning signals, requiring expert knowledge. While it is possible to train vision-based RL agents using only sparse rewards, additional challenges in exploration arise. We present a novel and efficient method to solve sparse-reward robot manipulation tasks from only image observations by utilizing a few demonstrations. First, we learn an embedded neural dynamics model from demonstration transitions and further fine-tune it with the replay buffer. Next, we reward the agents for staying close to the demonstrated trajectories using a distance metric defined in the embedding space. Finally, we use an off-policy, model-free vision RL algorithm to update the control policies. Our method achieves state-of-the-art sample efficiency in simulation and enables efficient training of a real Franka Emika Panda manipulator.

cs.RO↗

arXiv2023

Joint Learning of Reward Machines and Policies in Environments with Partially Known Semantics

We study the problem of reinforcement learning for a task encoded by a reward machine. The task is defined over a set of properties in the environment, called atomic propositions, and represented by Boolean variables. One unrealistic assumption commonly used in the literature is that the truth values of these propositions are accurately known. In real situations, however, these truth values are uncertain since they come from sensors that suffer from imperfections. At the same time, reward machines can be difficult to model explicitly, especially when they encode complicated tasks. We develop a reinforcement-learning algorithm that infers a reward machine that encodes the underlying task while learning how to execute it, despite the uncertainties of the propositions' truth values. In order to address such uncertainties, the algorithm maintains a probabilistic estimate about the truth value of the atomic propositions; it updates this estimate according to new sensory measurements that arrive from the exploration of the environment. Additionally, the algorithm maintains a hypothesis reward machine, which acts as an estimate of the reward machine that encodes the task to be learned. As the agent explores the environment, the algorithm updates the hypothesis reward machine according to the obtained rewards and the estimate of the atomic propositions' truth value. Finally, the algorithm uses a Q-learning procedure for the states of the hypothesis reward machine to determine the policy that accomplishes the task. We prove that the algorithm successfully infers the reward machine and asymptotically learns a policy that accomplishes the respective task.

cs.LG↗

arXiv2023

Forecaster-aided User Association and Load Balancing in Multi-band Mobile Networks

Cellular networks are becoming increasingly heterogeneous with higher base station (BS) densities and ever more frequency bands, making BS selection and band assignment key decisions in terms of rate and coverage. In this paper, we decompose the mobility-aware user association task into (i) forecasting of user rate and then (ii) convex utility maximization for user association accounting for the effects of BS load and handover overheads. Using a linear combination of normalized mean-squared error and normalized discounted cumulative gain as a novel loss function, a recurrent deep neural network is trained to reliably forecast the mobile users' future rates. Based on the forecast, the controller optimizes the association decisions to maximize the service rate-based network utility using our computationally efficient (speed up of 100x versus generic convex solver) algorithm based on the Frank-Wolfe method. Using an industry-grade network simulator developed by Meta, we show that the proposed model predictive control (MPC) approach improves the 5th percentile service rate by 3.5x compared to the traditional signal strength-based association, reduces the median number of handovers by 7x compared to a handover agnostic strategy, and achieves service rates close to a genie-aided scheme. Furthermore, our model-based approach is significantly more sample-efficient (needs 100x less training data) compared to model-free reinforcement learning (RL), and generalizes well across different user drop scenarios.

eess.SY↗

arXiv2022

Class-Aware Adversarial Transformers for Medical Image Segmentation

Transformers have made remarkable progress towards modeling long-range dependencies within the medical image analysis domain. However, current transformer-based models suffer from several disadvantages: (1) existing methods fail to capture the important features of the images due to the naive tokenization scheme; (2) the models suffer from information loss because they only consider single-scale feature representations; and (3) the segmentation label maps generated by the models are not accurate enough without considering rich semantic contexts and anatomical textures. In this work, we present CASTformer, a novel type of adversarial transformers, for 2D medical image segmentation. First, we take advantage of the pyramid structure to construct multi-scale representations and handle multi-scale variations. We then design a novel class-aware transformer module to better learn the discriminative regions of objects with semantic structures. Lastly, we utilize an adversarial training strategy that boosts segmentation accuracy and correspondingly allows a transformer-based discriminator to capture high-level semantically correlated contents and low-level anatomical features. Our experiments demonstrate that CASTformer dramatically outperforms previous state-of-the-art transformer-based approaches on three benchmarks, obtaining 2.54%-5.88% absolute improvements in Dice over previous models. Further qualitative experiments provide a more detailed picture of the model's inner workings, shed light on the challenges in improved transparency, and demonstrate that transfer learning can greatly improve performance and reduce the size of medical image datasets in training, making CASTformer a strong starting point for downstream medical image analysis tasks.

cs.CV↗

arXiv2022

Drift Reduced Navigation with Deep Explainable Features

Modern autonomous vehicles (AVs) often rely on vision, LIDAR, and even radar-based simultaneous localization and mapping (SLAM) frameworks for precise localization and navigation. However, modern SLAM frameworks often lead to unacceptably high levels of drift (i.e., localization error) when AVs observe few visually distinct features or encounter occlusions due to dynamic obstacles. This paper argues that minimizing drift must be a key desiderata in AV motion planning, which requires an AV to take active control decisions to move towards feature-rich regions while also minimizing conventional control cost. To do so, we first introduce a novel data-driven perception module that observes LIDAR point clouds and estimates which features/regions an AV must navigate towards for drift minimization. Then, we introduce an interpretable model predictive controller (MPC) that moves an AV toward such feature-rich regions while avoiding visual occlusions and gracefully trading off drift and control cost. Our experiments on challenging, dynamic scenarios in the state-of-the-art CARLA simulator indicate our method reduces drift up to 76.76% compared to benchmark approaches.

cs.RO↗

arXiv2022

Online Poisoning Attacks Against Data-Driven Predictive Control

Data-driven predictive control (DPC) is a feedback control method for systems with unknown dynamics. It repeatedly optimizes a system's future trajectories based on past input-output data. We develop a numerical method that computes poisoning attacks that inject additive perturbations to the online output data to change the trajectories optimized by DPC. This method is based on implicitly differentiating the solution map of the trajectory optimization in DPC. We demonstrate that the resulting attacks can cause an output tracking error one order of magnitude higher than random perturbations in numerical experiments.

eess.SY↗

arXiv2022

Task-Aware Network Coding Over Butterfly Network

Network coding allows distributed information sources such as sensors to efficiently compress and transmit data to distributed receivers across a bandwidth-limited network. Classical network coding is largely task-agnostic -- the coding schemes mainly aim to faithfully reconstruct data at the receivers, regardless of what ultimate task the received data is used for. In this paper, we analyze a new task-driven network coding problem, where distributed receivers pass transmitted data through machine learning (ML) tasks, which provides an opportunity to improve efficiency by transmitting salient task-relevant data representations. Specifically, we formulate a task-aware network coding problem over a butterfly network in real-coordinate space, where lossy analog compression through principal component analysis (PCA) can be applied. A lower bound for the total loss function for the formulated problem is given, and necessary and sufficient conditions for achieving this lower bound are also provided. We introduce ML algorithms to solve the problem in the general case, and our evaluation demonstrates the effectiveness of task-aware network coding.

cs.IT↗

arXiv2022

A Control Theoretic Approach to Infrastructure-Centric Blockchain Tokenomics

There are a multitude of Blockchain-based physical infrastructure systems, operating on a crypto-currency enabled token economy, where infrastructure suppliers are rewarded with tokens for enabling, validating, managing and/or securing the system. However, today's token economies are largely designed without infrastructure systems in mind, and often operate with a fixed token supply (e.g., Bitcoin). This paper argues that token economies for infrastructure networks should be structured differently - they should continually incentivize new suppliers to join the network to provide services and support to the ecosystem. As such, the associated token rewards should gracefully scale with the size of the decentralized system, but should be carefully balanced with consumer demand to manage inflation and be designed to ultimately reach an equilibrium. To achieve such an equilibrium, the decentralized token economy should be adaptable and controllable so that it maximizes the total utility of all users, such as achieving stable (overall non-inflationary) token economies. Our main contribution is to model infrastructure token economies as dynamical systems - the circulating token supply, price, and consumer demand change as a function of the payment to nodes and costs to consumers for infrastructure services. Crucially, this dynamical systems view enables us to leverage tools from mathematical control theory to optimize the overall decentralized network's performance. Moreover, our model extends easily to a Stackelberg game between the controller and the nodes, which we use for robust, strategic pricing. In short, we develop predictive, optimization-based controllers that outperform traditional algorithmic stablecoin heuristics by up to $2.4 \times$ in simulations based on real demand data from existing decentralized wireless networks.

cs.DC↗

arXiv2022

Task-aware Privacy Preservation for Multi-dimensional Data

Local differential privacy (LDP) can be adopted to anonymize richer user data attributes that will be input to sophisticated machine learning (ML) tasks. However, today's LDP approaches are largely task-agnostic and often lead to severe performance loss -- they simply inject noise to all data attributes according to a given privacy budget, regardless of what features are most relevant for the ultimate task. In this paper, we address how to significantly improve the ultimate task performance with multi-dimensional user data by considering a task-aware privacy preservation problem. The key idea is to use an encoder-decoder framework to learn (and anonymize) a task-relevant latent representation of user data. We obtain an analytical near-optimal solution for the linear setting with mean-squared error (MSE) task loss. We also provide an approximate solution through a gradient-based learning algorithm for general nonlinear cases. Extensive experiments demonstrate that our task-aware approach significantly improves ultimate task accuracy compared to standard benchmark LDP approaches with the same level of privacy guarantee.

cs.CR↗

arXiv2022

Dynamic Selection of Perception Models for Robotic Control

Robotic perception models, such as Deep Neural Networks (DNNs), are becoming more computationally intensive and there are several models being trained with accuracy and latency trade-offs. However, modern latency accuracy trade-offs largely report mean accuracy for single-step vision tasks, but there is little work showing which model to invoke for multi-step control tasks in robotics. The key challenge in a multi-step decision making is to make use of the right models at right times to accomplish the given task. That is, the accomplishment of the task with a minimum control cost and minimum perception time is a desideratum; this is known as the model selection problem. In this work, we precisely address this problem of invoking the correct sequence of perception models for multi-step control. In other words, we provide a provably optimal solution to the model selection problem by casting it as a multi-objective optimization problem balancing the control cost and perception time. The key insight obtained from our solution is how the variance of the perception models matters (not just the mean accuracy) for multi-step decision making, and to show how to use diverse perception models as a primitive for energy-efficient robotics. Further, we demonstrate our approach on a photo-realistic drone landing simulation using visual navigation in AirSim. Using our proposed policy, we achieved 38.04% lower control cost with 79.1% less perception time than other competing benchmarks.

cs.RO↗

arXiv2022

Learning Actions for Drift-Free Navigation in Highly Dynamic Scenes

We embark on a hitherto unreported problem of an autonomous robot (self-driving car) navigating in dynamic scenes in a manner that reduces its localization error and eventual cumulative drift or Absolute Trajectory Error, which is pronounced in such dynamic scenes. With the hugely popular Velodyne-16 3D LIDAR as the main sensing modality, and the accurate LIDAR-based Localization and Mapping algorithm, LOAM, as the state estimation framework, we show that in the absence of a navigation policy, drift rapidly accumulates in the presence of moving objects. To overcome this, we learn actions that lead to drift-minimized navigation through a suitable set of reward and penalty functions. We use Proximal Policy Optimization, a class of Deep Reinforcement Learning methods, to learn the actions that result in drift-minimized trajectories. We show by extensive comparisons on a variety of synthetic, yet photo-realistic scenes made available through the CARLA Simulator the superior performance of the proposed framework vis-a-vis methods that do not adopt such policies.

cs.RO↗

arXiv2021

Personalized Federated Learning of Driver Prediction Models for Autonomous Driving

Autonomous vehicles (AVs) must interact with a diverse set of human drivers in heterogeneous geographic areas. Ideally, fleets of AVs should share trajectory data to continually re-train and improve trajectory forecasting models from collective experience using cloud-based distributed learning. At the same time, these robots should ideally avoid uploading raw driver interaction data in order to protect proprietary policies (when sharing insights with other companies) or protect driver privacy from insurance companies. Federated learning (FL) is a popular mechanism to learn models in cloud servers from diverse users without divulging private local data. However, FL is often not robust -- it learns sub-optimal models when user data comes from highly heterogeneous distributions, which is a key hallmark of human-robot interactions. In this paper, we present a novel variant of personalized FL to specialize robust robot learning models to diverse user distributions. Our algorithm outperforms standard FL benchmarks by up to 2x in real user studies that we conducted where human-operated vehicles must gracefully merge lanes with simulated AVs in the standard CARLA and CARLO AV simulators.

cs.LG↗

arXiv2021

ML-EXray: Visibility into ML Deployment on the Edge

Benefiting from expanding cloud infrastructure, deep neural networks (DNNs) today have increasingly high performance when trained in the cloud. Researchers spend months of effort competing for an extra few percentage points of model accuracy. However, when these models are actually deployed on edge devices in practice, very often, the performance can abruptly drop over 10% without obvious reasons. The key challenge is that there is not much visibility into ML inference execution on edge devices, and very little awareness of potential issues during the edge deployment process. We present ML-EXray, an end-to-end framework, which provides visibility into layer-level details of the ML execution, and helps developers analyze and debug cloud-to-edge deployment issues. More often than not, the reason for sub-optimal edge performance does not only lie in the model itself, but every operation throughout the data flow and the deployment process. Evaluations show that ML-EXray can effectively catch deployment issues, such as pre-processing bugs, quantization issues, suboptimal kernels, etc. Using ML-EXray, users need to write less than 15 lines of code to fully examine the edge deployment pipeline. Eradicating these issues, ML-EXray can correct model performance by up to 30%, pinpoint error-prone layers, and guide users to optimize kernel execution latency by two orders of magnitude. Code and APIs will be released as an open-source multi-lingual instrumentation library and a Python deployment validation library.

cs.DC↗

arXiv2021

Data Sharing and Compression for Cooperative Networked Control

Sharing forecasts of network timeseries data, such as cellular or electricity load patterns, can improve independent control applications ranging from traffic scheduling to power generation. Typically, forecasts are designed without knowledge of a downstream controller's task objective, and thus simply optimize for mean prediction error. However, such task-agnostic representations are often too large to stream over a communication network and do not emphasize salient temporal features for cooperative control. This paper presents a solution to learn succinct, highly-compressed forecasts that are co-designed with a modular controller's task objective. Our simulations with real cellular, Internet-of-Things (IoT), and electricity load data show we can improve a model predictive controller's performance by at least $25\%$ while transmitting $80\%$ less data than the competing method. Further, we present theoretical compression results for a networked variant of the classical linear quadratic regulator (LQR) control problem.

cs.LG↗

arXiv2021

Combining Reinforcement Learning with Model Predictive Control for On-Ramp Merging

We consider the problem of designing an algorithm to allow a car to autonomously merge on to a highway from an on-ramp. Two broad classes of techniques have been proposed to solve motion planning problems in autonomous driving: Model Predictive Control (MPC) and Reinforcement Learning (RL). In this paper, we first establish the strengths and weaknesses of state-of-the-art MPC and RL-based techniques through simulations. We show that the performance of the RL agent is worse than that of the MPC solution from the perspective of safety and robustness to out-of-distribution traffic patterns, i.e., traffic patterns which were not seen by the RL agent during training. On the other hand, the performance of the RL agent is better than that of the MPC solution when it comes to efficiency and passenger comfort. We subsequently present an algorithm which blends the model-free RL agent with the MPC solution and show that it provides better trade-offs between all metrics -- passenger comfort, efficiency, crash rate and robustness.

cs.RO↗

arXiv2021

Interpretable Trade-offs Between Robot Task Accuracy and Compute Efficiency

A robot can invoke heterogeneous computation resources such as CPUs, cloud GPU servers, or even human computation for achieving a high-level goal. The problem of invoking an appropriate computation model so that it will successfully complete a task while keeping its compute and energy costs within a budget is called a model selection problem. In this paper, we present an optimal solution to the model selection problem with two compute models, the first being fast but less accurate, and the second being slow but more accurate. The main insight behind our solution is that a robot should invoke the slower compute model only when the benefits from the gain in accuracy outweigh the computational costs. We show that such cost-benefit analysis can be performed by leveraging the statistical correlation between the accuracy of fast and slow compute models. We demonstrate the broad applicability of our approach to diverse problems such as perception using neural networks and safe navigation of a simulated Mars rover.

cs.RO↗

arXiv2020

Sampling Training Data for Continual Learning Between Robots and the Cloud

Today's robotic fleets are increasingly measuring high-volume video and LIDAR sensory streams, which can be mined for valuable training data, such as rare scenes of road construction sites, to steadily improve robotic perception models. However, re-training perception models on growing volumes of rich sensory data in central compute servers (or the "cloud") places an enormous time and cost burden on network transfer, cloud storage, human annotation, and cloud computing resources. Hence, we introduce HarvestNet, an intelligent sampling algorithm that resides on-board a robot and reduces system bottlenecks by only storing rare, useful events to steadily improve perception models re-trained in the cloud. HarvestNet significantly improves the accuracy of machine-learning models on our novel dataset of road construction sites, field testing of self-driving cars, and streaming face recognition, while reducing cloud storage, dataset annotation time, and cloud compute time by between 65.7-81.3%. Further, it is between 1.05-2.58x more accurate than baseline algorithms and scalably runs on embedded deep learning hardware. We provide a suite of compute-efficient perception models for the Google Edge Tensor Processing Unit (TPU), an extended technical report, and a novel video dataset to the research community at https://sites.google.com/view/harvestnet.

cs.RO↗