arXiv subjects

Prajwal Koirala

Publications and source records attributed to Prajwal Koirala.

9 recordsLinked to original sources

arXiv2026

LexiSafe: Offline Safe Reinforcement Learning with Lexicographic Safety-Reward Hierarchy

Offline safe reinforcement learning (RL) is increasingly important for cyber-physical systems (CPS), where safety violations during training are unacceptable and only pre-collected data are available. Existing offline safe RL methods typically balance reward-safety tradeoffs through constraint relaxation or joint optimization, but they often lack structural mechanisms to prevent safety drift. We propose LexiSafe, a lexicographic offline RL framework designed to preserve safety-aligned behavior. We first develop LexiSafe-SC, a single-cost formulation for standard offline safe RL, and derive safety-violation and performance-suboptimality bounds that together yield sample-complexity guarantees. We then extend the framework to hierarchical safety requirements with LexiSafe-MC, which supports multiple safety costs and admits its own sample-complexity analysis. Empirically, LexiSafe demonstrates reduced safety violations and improved task performance compared to constrained offline baselines. By unifying lexicographic prioritization with structural bias, LexiSafe offers a practical and theoretically grounded approach for safety-critical CPS decision-making.

cs.LG↗

arXiv2026

LCLA: Language-Conditioned Latent Alignment for Vision-Language Navigation

We propose LCLA (Language-Conditioned Latent Alignment), a framework for vision-language navigation that learns modular perception-action interfaces by aligning sensory observations to a latent representation of an expert policy. The expert is first trained with privileged state information, inducing a latent space sufficient for control, after which its latent interface and action head are frozen. A lightweight adapter is then trained to map raw visual-language observations, via a frozen vision-language model, into the expert's latent space, reducing the problem of visuomotor learning to supervised latent alignment rather than end-to-end policy optimization. This decoupling enforces a stable contract between perception and control, enabling expert behavior to be reused across sensing modalities and environmental variations. We instantiate LCLA and evaluate it on a vision-language indoor navigation task, where aligned latent spaces yield strong in-distribution performance and robust zero-shot generalization to unseen environments, lighting conditions, and viewpoints while remaining lightweight at inference time.

cs.RO↗

arXiv2025

Neural CDEs as Correctors for Learned Time Series Models

Learned time-series models, whether continuous or discrete, are widely used for forecasting the states of dynamical systems but suffer from error accumulation in multi-step forecasts. To address this issue, we propose a Predictor-Corrector framework in which the Predictor is a learned time-series model that generates multi-step forecasts and the Corrector is a neural controlled differential equation that corrects the forecast errors. The Corrector works with irregularly sampled time series and is compatible with both continuous- and discrete-time Predictors. We further introduce two regularization strategies that improve the Corrector's extrapolation performance and accelerate its training. We also provide theoretical guarantees on the stability and convergence of the proposed framework. Experiments on synthetic, physics-based, and real-world datasets show that the proposed framework consistently improves forecasting performance across diverse Predictors, including neural ordinary differential equations, ContiFormer, and DLinear, demonstrating its predictor-agnostic nature.

cs.LG↗

arXiv2025

Flow-Based Single-Step Completion for Efficient and Expressive Policy Learning

Generative models such as diffusion and flow-matching offer expressive policies for offline reinforcement learning (RL) by capturing rich, multimodal action distributions, but their iterative sampling introduces high inference costs and training instability due to gradient propagation across sampling steps. We propose the Single-Step Completion Policy (SSCP), a generative policy trained with an augmented flow-matching objective to predict direct completion vectors from intermediate flow samples, enabling accurate, one-shot action generation. In an off-policy actor-critic framework, SSCP combines the expressiveness of generative models with the training and inference efficiency of unimodal policies, without requiring long backpropagation chains. Our method scales effectively to offline, offline-to-online, and online RL settings, offering substantial gains in speed and adaptability over diffusion-based baselines. We further extend SSCP to goal-conditioned RL, enabling flat policies to exploit subgoal structures without explicit hierarchical inference. SSCP achieves strong results across standard offline RL and behavior cloning benchmarks, positioning it as a versatile, expressive, and efficient framework for deep RL and sequential decision-making. The code is available at https://github.com/PrajwalKoirala/SSCP-Single-Step-Completion-Policy.

cs.LG↗

arXiv2025

Can Pretrained Vision-Language Embeddings Alone Guide Robot Navigation?

Foundation models have revolutionized robotics by providing rich semantic representations without task-specific training. While many approaches integrate pretrained vision-language models (VLMs) with specialized navigation architectures, the fundamental question remains: can these pretrained embeddings alone successfully guide navigation without additional fine-tuning or specialized modules? We present a minimalist framework that decouples this question by training a behavior cloning policy directly on frozen vision-language embeddings from demonstrations collected by a privileged expert. Our approach achieves a 74% success rate in navigation to language-specified targets, compared to 100% for the state-aware expert, though requiring 3.2 times more steps on average. This performance gap reveals that pretrained embeddings effectively support basic language grounding but struggle with long-horizon planning and spatial reasoning. By providing this empirical baseline, we highlight both the capabilities and limitations of using foundation models as drop-in representations for embodied tasks, offering critical insights for robotics researchers facing practical design tradeoffs between system complexity and performance in resource-constrained scenarios. Our code is available at https://github.com/oadamharoon/text2nav

cs.RO↗

arXiv2024

FAWAC: Feasibility Informed Advantage Weighted Regression for Persistent Safety in Offline Reinforcement Learning

Safe offline reinforcement learning aims to learn policies that maximize cumulative rewards while adhering to safety constraints, using only offline data for training. A key challenge is balancing safety and performance, particularly when the policy encounters out-of-distribution (OOD) states and actions, which can lead to safety violations or overly conservative behavior during deployment. To address these challenges, we introduce Feasibility Informed Advantage Weighted Actor-Critic (FAWAC), a method that prioritizes persistent safety in constrained Markov decision processes (CMDPs). FAWAC formulates policy optimization with feasibility conditions derived specifically for offline datasets, enabling safe policy updates in non-parametric policy space, followed by projection into parametric space for constrained actor training. By incorporating a cost-advantage term into Advantage Weighted Regression (AWR), FAWAC ensures that the safety constraints are respected while maximizing performance. Additionally, we propose a strategy to address a more challenging class of problems that involves tempting datasets where trajectories are predominantly high-rewarded but unsafe. Empirical evaluations on standard benchmarks demonstrate that FAWAC achieves strong results, effectively balancing safety and performance in learning policies from the static datasets.

cs.LG↗

arXiv2024

Latent Safety-Constrained Policy Approach for Safe Offline Reinforcement Learning

In safe offline reinforcement learning (RL), the objective is to develop a policy that maximizes cumulative rewards while strictly adhering to safety constraints, utilizing only offline data. Traditional methods often face difficulties in balancing these constraints, leading to either diminished performance or increased safety risks. We address these issues with a novel approach that begins by learning a conservatively safe policy through the use of Conditional Variational Autoencoders, which model the latent safety constraints. Subsequently, we frame this as a Constrained Reward-Return Maximization problem, wherein the policy aims to optimize rewards while complying with the inferred latent safety constraints. This is achieved by training an encoder with a reward-Advantage Weighted Regression objective within the latent constraint space. Our methodology is supported by theoretical analysis, including bounds on policy performance and sample complexity. Extensive empirical evaluation on benchmark datasets, including challenging autonomous driving scenarios, demonstrates that our approach not only maintains safety compliance but also excels in cumulative reward optimization, surpassing existing methods. Additional visualizations provide further insights into the effectiveness and underlying mechanisms of our approach.

cs.LG↗

arXiv2024

F1tenth Autonomous Racing With Offline Reinforcement Learning Methods

Autonomous racing serves as a critical platform for evaluating automated driving systems and enhancing vehicle mobility intelligence. This work investigates offline reinforcement learning methods to train agents within the dynamic F1tenth racing environment. The study begins by exploring the challenges of online training in the Austria race track environment, where agents consistently fail to complete the laps. Consequently, this research pivots towards an offline strategy, leveraging `expert' demonstration dataset to facilitate agent training. A waypoint-based suboptimal controller is developed to gather data with successful lap episodes. This data is then employed to train offline learning-based algorithms, with a subsequent analysis of the agents' cross-track performance, evaluating their zero-shot transferability from seen to unseen scenarios and their capacity to adapt to changes in environment dynamics. Beyond mere algorithm benchmarking in autonomous racing scenarios, this study also introduces and describes the machinery of our return-conditioned decision tree-based policy, comparing its performance with methods that employ fully connected neural networks, Transformers, and Diffusion Policies and highlighting some insights into method selection for training autonomous agents in driving interactions.

cs.RO↗

arXiv2024

Solving Offline Reinforcement Learning with Decision Tree Regression

This study presents a novel approach to addressing offline reinforcement learning (RL) problems by reframing them as regression tasks that can be effectively solved using Decision Trees. Mainly, we introduce two distinct frameworks: return-conditioned and return-weighted decision tree policies (RCDTP and RWDTP), both of which achieve notable speed in agent training as well as inference, with training typically lasting less than a few minutes. Despite the simplification inherent in this reformulated approach to offline RL, our agents demonstrate performance that is at least on par with the established methods. We evaluate our methods on D4RL datasets for locomotion and manipulation, as well as other robotic tasks involving wheeled and flying robots. Additionally, we assess performance in delayed/sparse reward scenarios and highlight the explainability of these policies through action distribution and feature importance.

cs.LG↗