Gokul Puthumanaillam

Gokul Puthumanaillam, Hao Jiang, Ruben Hernandez et al.

Diffusion-based trajectory planners can synthesize rich, multimodal robot motions, but their iterative denoising makes online planning and control prohibitively slow. Existing accelerations either modify the sampler or compress the network--sacrificing plan quality or requiring retraining without accounting for downstream control risk. We address the problem of making diffusion-based trajectory planners fast enough for real-time robot use without retraining the model or sacrificing trajectory quality, and in a way that works across diverse state-space diffusion architectures. Our key insight is that diffusion trajectory planners expose two signals we can exploit: a cheap probe of how their internal trajectory representation changes across steps, and analytic coefficients that describe how denoiser errors affect the sampler's state update. By calibrating the first signal against the second on offline runs, we obtain a per-step score that upper-bounds how far the final trajectory can deviate when we reuse a cached denoiser output, and we treat this bound as an uncertainty budget that we can spend over the denoising process. Building on this insight, we present Muninn, a training-free caching wrapper that tracks this uncertainty budget during sampling and, at each diffusion step, chooses between reusing a cached denoiser output when the predicted deviation is small and recomputing the denoiser when it is not. Across standard benchmarks Muninn delivers up to 4.6x wall-clock speedups across several trajectory diffusion models by reducing denoiser evaluations, while preserving task performance and safety metrics. Muninn further certifies that cached rollouts remain within a specified distance of their full-compute counterparts, and we validate these gains in real-time closed-loop navigation and manipulation hardware deployments. Project page: https://github.com/gokulp01/Muninn.

Gokul Puthumanaillam, Melkior Ornik

Diffusion-based trajectory planners can synthesize rich, multimodal action sequences for offline reinforcement learning, but their iterative denoising incurs substantial inference-time cost, making closed-loop planning slow under tight compute budgets. We study the problem of achieving diffusion-like trajectory planning behavior with one-step inference, while retaining the ability to sample diverse candidate plans and condition on the current state in a receding-horizon control loop. Our key observation is that conditional trajectory generation fails under naïve distribution-matching objectives when the similarity measure used to align generated trajectories with the dataset is dominated by unconstrained future dimensions. In practice, this causes attraction toward average trajectories, collapses action diversity, and yields near-static behavior. Our key insight is that conditional generative planning requires a conditioning-aware notion of neighborhood: trajectory updates should be computed using distances in a compact key space that reflects the condition, while still applying updates in the full trajectory space. Building on this, we introduce Keyed Drifting Policies (KDP), a one-step trajectory generator trained with a drift-field objective that attracts generated trajectories toward condition-matched dataset windows and repels them from nearby generated samples, using a stop-gradient drifted target to amortize iterative refinement into training. At inference, the resulting policy produces a full trajectory window in a single forward pass. Across standard RL benchmarks and real-time hardware deployments, KDP achieves strong performance with one-step inference and substantially lower planning latency than diffusion sampling. Project website, code and videos: https://keyed-drifting.github.io/

Gokul Puthumanaillam, Manav Vora, Pranay Thangeda et al.

This paper examines the challenges associated with achieving life-long superalignment in AI systems, particularly large language models (LLMs). Superalignment is a theoretical framework that aspires to ensure that superintelligent AI systems act in accordance with human values and goals. Despite its promising vision, we argue that achieving superalignment requires substantial changes in the current LLM architectures due to their inherent limitations in comprehending and adapting to the dynamic nature of these human ethics and evolving global scenarios. We dissect the challenges of encoding an ever-changing spectrum of human values into LLMs, highlighting the discrepancies between static AI models and the dynamic nature of human societies. To illustrate these challenges, we analyze two distinct examples: one demonstrates a qualitative shift in human values, while the other presents a quantifiable change. Through these examples, we illustrate how LLMs, constrained by their training data, fail to align with contemporary human values and scenarios. The paper concludes by exploring potential strategies to address and possibly mitigate these alignment discrepancies, suggesting a path forward in the pursuit of more adaptable and responsive AI systems.

Negin Musavi, Gokul Puthumanaillam, Ruben Hernandez et al.

Robots operating in non-stationary environments must continually adapt their policies as the dynamics drift, but onboard energy and compute budgets cap how often a full state estimation and re-planning step can be performed. This raises a question: \emph{when}, along a horizon, should a robot spend its limited budget? We formulate this problem in time-varying Markov decision processes (TVMDPs) with a known bound on the rate of transition drift. We model execution as a \emph{skip-update} scheme in which, at chosen update times, the agent estimates the transition kernel by maximum likelihood and computes a finite-horizon policy, and between updates reuses this policy under a propagated state estimate. We analyze the dynamic regret of this scheme and show how it grows during skip intervals in terms of the properties of the TVMDP and the skip lengths; the resulting bound answers the opening question via an online, regret-guided update rule that allocates the budget adaptively. We evaluate the rule in a simulated Mars-rover navigation task with time-varying slip dynamics and on a Crazyflie quadrotor in indoor obstacle fields. Adaptive allocation outperforms other budgeted baselines.

Gokul Puthumanaillam, Xiangyu Liu, Negar Mehr et al.

Optimal decision-making presents a significant challenge for autonomous systems operating in uncertain, stochastic and time-varying environments. Environmental variability over time can significantly impact the system's optimal decision making strategy for mission completion. To model such environments, our work combines the previous notion of Time-Varying Markov Decision Processes (TVMDP) with partial observability and introduces Time-Varying Partially Observable Markov Decision Processes (TV-POMDP). We propose a two-pronged approach to accurately estimate and plan within the TV-POMDP: 1) Memory Prioritized State Estimation (MPSE), which leverages weighted memory to provide more accurate time-varying transition estimates; and 2) an MPSE-integrated planning strategy that optimizes long-term rewards while accounting for temporal constraint. We validate the proposed framework and algorithms using simulations and hardware, with robots exploring a partially observable, time-varying environments. Our results demonstrate superior performance over standard methods, highlighting the framework's effectiveness in stochastic, uncertain, time-varying domains.

Bharawee Phoompho, Gokul Puthumanaillam, Yan Miao et al.

Reliable trajectory planning under partial observability depends not only on computing a feasible geometric path, but also on whether the robot receives informative observations while executing that trajectory. Existing approaches usually keep the environment fixed and adapt the robot through belief-space planning, active localization, or added sensing, often incurring costly uncertainty propagation and brittle behavior in observation-poor regions. We flip this perspective and address the largely open problem of \emph{task-aware environment augmentation}: given a mapped environment, a planned task trajectory, and a small budget of visual fiducial markers, where should the environment be augmented so that the planned trajectory can be executed reliably under uncertainty? Our key observation is that useful marker layouts are defined by the localization support they provide along the task trajectory: a small number of well-timed observations can be sufficient to prevent uncertainty from accumulating in regions where state-estimation error would otherwise compromise control. Building on this observation, we present \tbp{SCoDA}, $\textbf{S}$hielded $\textbf{Co}$nditional $\textbf{D}$iffusion for Environment $\textbf{A}$ugmentation. \tbp{SCoDA} learns a conditional distribution over high-performing fiducial layouts from data, using the environment, planned trajectory, disturbance context, and desired execution profile as conditioning. Its shielded sampler reasons over where along the planned execution pose corrections should occur, and steers this distribution toward task-relevant, finite-budget augmentations. Across simulated benchmarks and hardware deployments, we show that \tbp{SCoDA} improves trajectory execution reliability and completion time over strong baselines. Code, models and dataset available at: \hyperlink{scoda-diffusion.github.io}{https://scoda-diffusion.github.io/}

Gokul Puthumanaillam, Vardhan Dongre, Pranay Thangeda et al.

Vision-language-action (VLA) policies bring natural language into closed-loop robot control, enabling robots to execute manipulation tasks directly from text instructions. The same interface gives text a recurring role in control because the prompt is reused at every replanning step, and each prompt-conditioned action changes the future observations on which the policy acts. Existing VLA attacks study adversarial prompts that elicit targeted low-level actions or make such actions persist across changing images. We identify a stronger trajectory-level failure mode: a prompt that still $\textit{appears}$ to specify the intended task but redirects the final physical outcome. We mathematically formalize this setting as $\textit{command-preserving trajectory redirection}$, a prompt-only threat model in which the attacker chooses one prompt before the episode, all policy and environment components remain fixed, and the prompt must stay close to the benign instruction while omitting target words and correction language. To find such prompts, we introduce an on-policy prompt search method that uses rollouts to discover perturbations whose closed-loop behavior tracks a target task while satisfying the command-preserving constraints. Experiments in simulation and on hardware show that near-benign prompt perturbations can redirect VLA rollouts to attacker-specified targets. These results expose a trajectory-level vulnerability in VLA instruction grounding: text that appears to preserve the intended command can still give an adversary control over the robot's final physical outcome. Project website: https://vla-redirection-attack.github.io/

Gokul Puthumanaillam, Manav Vora, Melkior Ornik

Optimal decision-making for trajectory tracking in partially observable, stochastic environments where the number of active localization updates -- the process by which the agent obtains its true state information from the sensors -- are limited, presents a significant challenge. Traditional methods often struggle to balance resource conservation, accurate state estimation and precise tracking, resulting in suboptimal performance. This problem is particularly pronounced in environments with large action spaces, where the need for frequent, accurate state data is paramount, yet the capacity for active localization updates is restricted by external limitations. This paper introduces ComTraQ-MPC, a novel framework that combines Deep Q-Networks (DQN) and Model Predictive Control (MPC) to optimize trajectory tracking with constrained active localization updates. The meta-trained DQN ensures adaptive active localization scheduling, while the MPC leverages available state information to improve tracking. The central contribution of this work is their reciprocal interaction: DQN's update decisions inform MPC's control strategy, and MPC's outcomes refine DQN's learning, creating a cohesive, adaptive system. Empirical evaluations in simulated and real-world settings demonstrate that ComTraQ-MPC significantly enhances operational efficiency and accuracy, providing a generalizable and approximately optimal solution for trajectory tracking in complex partially observable environments.

Gokul Puthumanaillam, Paulo Padrao, Jose Fuentes et al.

Robots navigating complex environments must manage uncertainty from sensor noise, environmental changes, and incomplete information, with different tasks requiring varying levels of precision in different areas. For example, precise localization may be crucial near obstacles but less critical in open spaces. We present GUIDE (Generalized Uncertainty Integration for Decision-Making and Execution), a framework that integrates these task-specific requirements into navigation policies via Task-Specific Uncertainty Maps (TSUMs). By assigning acceptable uncertainty levels to different locations, TSUMs enable robots to adapt uncertainty management based on context. When combined with reinforcement learning, GUIDE learns policies that balance task completion and uncertainty management without extensive reward engineering. Real-world tests show significant performance gains over methods lacking task-specific uncertainty awareness.

Gokul Puthumanaillam, Jae Hyuk Song, Nurzhan Yesmagambet et al.

Autonomous agents operating in adversarial scenarios face a fundamental challenge: while they may know their adversaries' high-level objectives, such as reaching specific destinations within time constraints, the exact policies these adversaries will employ remain unknown. Traditional approaches address this challenge by treating the adversary's state as a partially observable element, leading to a formulation as a Partially Observable Markov Decision Process (POMDP). However, the induced belief-space dynamics in a POMDP require knowledge of the system's transition dynamics, which, in this case, depend on the adversary's unknown policy. Our key observation is that while an adversary's exact policy is unknown, their behavior is necessarily constrained by their mission objectives and the physical environment, allowing us to characterize the space of possible behaviors without assuming specific policies. In this paper, we develop Task-Aware Behavior Fields (TAB-Fields), a representation that captures adversary state distributions over time by computing the most unbiased probability distribution consistent with known constraints. We construct TAB-Fields by solving a constrained optimization problem that minimizes additional assumptions about adversary behavior beyond mission and environmental requirements. We integrate TAB-Fields with standard planning algorithms by introducing TAB-conditioned POMCP, an adaptation of Partially Observable Monte Carlo Planning. Through experiments in simulation with underwater robots and hardware implementations with ground robots, we demonstrate that our approach achieves superior performance compared to baselines that either assume specific adversary policies or neglect mission constraints altogether. Evaluation videos and code are available at https://tab-fields.github.io.

Gokul Puthumanaillam, Paulo Padrao, Jose Fuentes et al.

Autonomous vehicles performing navigation tasks in complex environments face significant challenges due to uncertainty in state estimation. In many scenarios, such as stealth operations or resource-constrained settings, accessing high-precision localization comes at a significant cost, forcing robots to rely primarily on less precise state estimates. Our key observation is that different tasks require varying levels of precision in different regions: a robot navigating a crowded space might need precise localization near obstacles but can operate effectively with less precision elsewhere. In this paper, we present a planning method for integrating task-specific uncertainty requirements directly into navigation policies. We introduce Task-Specific Uncertainty Maps (TSUMs), which abstract the acceptable levels of state estimation uncertainty across different regions. TSUMs align task requirements and environmental features using a shared representation space, generated via a domain-adapted encoder. Using TSUMs, we propose Generalized Uncertainty Integration for Decision-Making and Execution (GUIDE), a policy conditioning framework that incorporates these uncertainty requirements into robot decision-making. We find that TSUMs provide an effective way to abstract task-specific uncertainty requirements, and conditioning policies on TSUMs enables the robot to reason about the context-dependent value of certainty and adapt its behavior accordingly. We show how integrating GUIDE into reinforcement learning frameworks allows the agent to learn navigation policies that effectively balance task completion and uncertainty management without explicit reward engineering. We evaluate GUIDE on various real-world robotic navigation tasks and find that it demonstrates significant improvement in task completion rates compared to baseline methods that do not explicitly consider task-specific uncertainty.

Manav Vora, Gokul Puthumanaillam, Hiroyasu Tsukamoto et al.

Communication can improve coordination in partially observed multi-agent reinforcement learning (MARL), but learning \emph{when} and \emph{who} to communicate with requires choosing among many possible sender-recipient pairs, and the effect of any single message on future reward is hard to isolate. We introduce \textbf{SCoUT} (\textbf{S}calable \textbf{Co}mmunication via \textbf{U}tility-guided \textbf{T}emporal grouping), which addresses both these challenges via temporal and agent abstraction within traditional MARL. During training, SCoUT resamples \textit{soft} agent groups every \(K\) environment steps (macro-steps) via Gumbel-Softmax; these groups are latent clusters that induce an affinity used as a differentiable prior over recipients. Using the same assignments, a group-aware critic predicts values for each agent group and maps them to per-agent baselines through the same soft assignments, reducing critic complexity and variance. Each agent is trained with a three-headed policy: environment action, send decision, and recipient selection. To obtain precise communication learning signals, we derive counterfactual communication advantages by analytically removing each sender's contribution from the recipient's aggregated messages. This counterfactual computation enables precise credit assignment for both send and recipient-selection decisions. At execution time, all centralized training components are discarded and only the per-agent policy is run, preserving decentralized execution. Project website, videos and code: \hyperlink{https://scout-comm.github.io/}{https://scout-comm.github.io/}