arXiv subjects

Melkior Ornik

Publications and source records attributed to Melkior Ornik.

At least 73 records · Page 4Linked to original sources

arXiv2024

ComTraQ-MPC: Meta-Trained DQN-MPC Integration for Trajectory Tracking with Limited Active Localization Updates

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.

cs.RO↗

arXiv2024

How Much Reserve Fuel: Quantifying the Maximal Energy Cost of System Disturbances

Motivated by the design question of additional fuel needed to complete a task in an uncertain environment, this paper introduces metrics to quantify the maximal additional energy used by a control system in the presence of bounded disturbances when compared to a nominal, disturbance-free system. In particular, we consider the task of finite-time stabilization for a linear time-invariant system. We first derive the nominal energy required to achieve this task in a disturbance-free system, and then the worst-case energy over all feasible disturbances. The latter leads to an optimal control problem with a least-squares solution, and then an infinite-dimensional optimization problem where we derive an upper bound on the solution. The comparison of these energies is accomplished using additive and multiplicative metrics, and we derive analytical bounds on these metrics. Simulation examples on an ADMIRE fighter jet model demonstrate the practicability of these metrics, and their variation with the task hardness, a combination of the distance of the initial condition from the origin and the task completion time.

eess.SY↗

arXiv2024

Few-shot Scooping Under Domain Shift via Simulated Maximal Deployment Gaps

Autonomous lander missions on extraterrestrial bodies need to sample granular materials while coping with domain shifts, even when sampling strategies are extensively tuned on Earth. To tackle this challenge, this paper studies the few-shot scooping problem and proposes a vision-based adaptive scooping strategy that uses the deep kernel Gaussian process method trained with a novel meta-training strategy to learn online from very limited experience on out-of-distribution target terrains. Our Deep Kernel Calibration with Maximal Deployment Gaps (kCMD) strategy explicitly trains a deep kernel model to adapt to large domain shifts by creating simulated maximal deployment gaps from an offline training dataset and training models to overcome these deployment gaps during training. Employed in a Bayesian Optimization sequential decision-making framework, the proposed method allows the robot to perform high-quality scooping actions on out-of-distribution terrains after a few attempts, significantly outperforming non-adaptive methods proposed in the excavation literature as well as other state-of-the-art meta-learning methods. The proposed method also demonstrates zero-shot transfer capability, successfully adapting to the NASA OWLAT platform, which serves as a state-of-the-art simulator for potential future planetary missions. These results demonstrate the potential of training deep models with simulated deployment gaps for more generalizable meta-learning in high-capacity models. Furthermore, they highlight the promise of our method in autonomous lander sampling missions by enabling landers to overcome the deployment gap between Earth and extraterrestrial bodies.

cs.RO↗

arXiv2024

Learning to Turn: Diffusion Imitation for Robust Row Turning in Under-Canopy Robots

Under-canopy agricultural robots require robust navigation capabilities to enable full autonomy but struggle with tight row turning between crop rows due to degraded GPS reception, visual aliasing, occlusion, and complex vehicle dynamics. We propose an imitation learning approach using diffusion policies to learn row turning behaviors from demonstrations provided by human operators or privileged controllers. Simulation experiments in a corn field environment show potential in learning this task with only visual observations and velocity states. However, challenges remain in maintaining control within rows and handling varied initial conditions, highlighting areas for future improvement.

cs.RO↗

arXiv2024

A Moral Imperative: The Need for Continual Superalignment of Large Language Models

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.

cs.CY↗

arXiv2024

Weathering Ongoing Uncertainty: Learning and Planning in a Time-Varying Partially Observable Environment

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.

cs.RO↗

arXiv2024

Identity Concealment Games: How I Learned to Stop Revealing and Love the Coincidences

In an adversarial environment, a hostile player performing a task may behave like a non-hostile one in order not to reveal its identity to an opponent. To model such a scenario, we define identity concealment games: zero-sum stochastic reachability games with a zero-sum objective of identity concealment. To measure the identity concealment of the player, we introduce the notion of an average player. The average player's policy represents the expected behavior of a non-hostile player. We show that there exists an equilibrium policy pair for every identity concealment game and give the optimality equations to synthesize an equilibrium policy pair. If the player's opponent follows a non-equilibrium policy, the player can hide its identity better. For this reason, we study how the hostile player may learn the opponent's policy. Since learning via exploration policies would quickly reveal the hostile player's identity to the opponent, we consider the problem of learning a near-optimal policy for the hostile player using the game runs collected under the average player's policy. Consequently, we propose an algorithm that provably learns a near-optimal policy and give an upper bound on the number of sample runs to be collected.

cs.GT↗

arXiv2024

Losing Control of your Network? Try Resilience Theory

Resilience of cyber-physical networks to unexpected failures is a critical need widely recognized across domains. For instance, power grids, telecommunication networks, transportation infrastructures and water treatment systems have all been subject to disruptive malfunctions and catastrophic cyber-attacks. Following such adverse events, we investigate scenarios where a node of a linear network suffers a loss of control authority over some of its actuators. These actuators are not following the controller's commands and are instead producing undesirable outputs. The repercussions of such a loss of control can propagate and destabilize the whole network despite the malfunction occurring at a single node. To assess system vulnerability, we establish resilience conditions for networks with a subsystem enduring a loss of control authority over some of its actuators. Furthermore, we quantify the destabilizing impact on the overall network when such a malfunction perturbs a nonresilient subsystem. We illustrate our resilience conditions on two academic examples, on an islanded microgrid, and on the linearized IEEE 39-bus system.

eess.SY↗

arXiv2023

Learning and Autonomy for Extraterrestrial Terrain Sampling: An Experience Report from OWLAT Deployment

Extraterrestrial autonomous lander missions increasingly demand adaptive capabilities to handle the unpredictable and diverse nature of the terrain. This paper discusses the deployment of a Deep Meta-Learning with Controlled Deployment Gaps (CoDeGa) trained model for terrain scooping tasks in Ocean Worlds Lander Autonomy Testbed (OWLAT) at NASA Jet Propulsion Laboratory. The CoDeGa-powered scooping strategy is designed to adapt to novel terrains, selecting scooping actions based on the available RGB-D image data and limited experience. The paper presents our experiences with transferring the scooping framework with CoDeGa-trained model from a low-fidelity testbed to the high-fidelity OWLAT testbed. Additionally, it validates the method's performance in novel, realistic environments, and shares the lessons learned from deploying learning-based autonomy algorithms for space exploration. Experimental results from OWLAT substantiate the efficacy of CoDeGa in rapidly adapting to unfamiliar terrains and effectively making autonomous decisions under considerable domain shifts, thereby endorsing its potential utility in future extraterrestrial missions.

cs.RO↗

arXiv2023

Few-shot Adaptation for Manipulating Granular Materials Under Domain Shift

Autonomous lander missions on extraterrestrial bodies will need to sample granular material while coping with domain shift, no matter how well a sampling strategy is tuned on Earth. This paper proposes an adaptive scooping strategy that uses deep Gaussian process method trained with meta-learning to learn on-line from very limited experience on the target terrains. It introduces a novel meta-training approach, Deep Meta-Learning with Controlled Deployment Gaps (CoDeGa), that explicitly trains the deep kernel to predict scooping volume robustly under large domain shifts. Employed in a Bayesian Optimization sequential decision-making framework, the proposed method allows the robot to use vision and very little on-line experience to achieve high-quality scooping actions on out-of-distribution terrains, significantly outperforming non-adaptive methods proposed in the excavation literature as well as other state-of-the-art meta-learning methods. Moreover, a dataset of 6,700 executed scoops collected on a diverse set of materials, terrain topography, and compositions is made available for future research in granular material manipulation and meta-learning.

cs.RO↗

arXiv2023

Viability under Degraded Control Authority

In this work, we solve the problem of quantifying and mitigating control authority degradation in real time. Here, our target systems are controlled nonlinear affine-in-control evolution equations with finite control input and finite- or infinite-dimensional state. We consider two cases of control input degradation: finitely many affine maps acting on unknown disjoint subsets of the inputs and general Lipschitz continuous maps. These degradation modes are encountered in practice due to actuator wear and tear, hard locks on actuator ranges due to over-excitation, as well as more general changes in the control allocation dynamics. We derive sufficient conditions for identifiability of control authority degradation, and propose a novel real-time algorithm for identifying or approximating control degradation modes. We demonstrate our method on a nonlinear distributed parameter system, namely a one-dimensional heat equation with a velocity-controlled moveable heat source, motivated by autonomous energy-based surgery.

eess.SY↗

arXiv2023

Identifying Single-Input Linear System Dynamics from Reachable Sets

This paper is concerned with identifying linear system dynamics without the knowledge of individual system trajectories, but from the knowledge of the system's reachable sets observed at different times. Motivated by a scenario where the reachable sets are known from partially transparent manufacturer specifications or observations of the collective behavior of adversarial agents, we aim to utilize such sets to determine the unknown system's dynamics. This paper has two contributions. Firstly, we show that the sequence of the system's reachable sets can be used to uniquely determine the system's dynamics for asymmetric input sets under some generic assumptions, regardless of the system's dimensions. We also prove the same property holds up to a sign change for two-dimensional systems where the input set is symmetric around zero. Secondly, we present an algorithm to determine these dynamics. We apply and verify the developed theory and algorithms on an unknown band-pass filter circuit solely provided the unknown system's reachable sets over a finite observation period.

eess.SY↗

arXiv2023

Delayed resilient trajectory tracking after partial loss of control authority over actuators

After the loss of control authority over thrusters of the Nauka module, the International Space Station lost attitude control for 45 minutes with potentially disastrous consequences. Motivated by a scenario of orbital inspection, we consider a similar malfunction occurring to the inspector satellite and investigate whether its mission can still be safely fulfilled. While a natural approach is to counteract in real-time the uncontrolled and undesirable thrust with the remaining controlled thrusters, vehicles are often subject to actuation delays hindering this approach. Instead, we extend resilience theory to systems suffering from actuation delay and build a resilient trajectory tracking controller with stability guarantees relying on a state predictor. We demonstrate that this controller can track accurately the reference trajectory of the inspection mission despite the actuation delay and the loss of control authority over one of the thrusters.

eess.SY↗

arXiv2023

Multi-agent Multi-target Path Planning in Markov Decision Processes

Missions for autonomous systems often require agents to visit multiple targets in complex operating conditions. This work considers the problem of visiting a set of targets in minimum time by a team of non-communicating agents in a Markov decision process (MDP). The single-agent problem is at least NP-complete by reducing it to a Hamiltonian path problem. We first discuss an optimal algorithm based on Bellman's optimality equation that is exponential in the number of target states. Then, we trade-off optimality for time complexity by presenting a suboptimal algorithm that is polynomial at each time step. We prove that the proposed algorithm generates optimal policies for certain classes of MDPs. Extending our procedure to the multi-agent case, we propose a target partitioning algorithm that approximately minimizes the expected time to visit the targets. We prove that our algorithm generates optimal partitions for clustered target scenarios. We present the performance of our algorithms on random MDPs and gridworld environments inspired by ocean dynamics. We show that our algorithms are much faster than the optimal procedure and more optimal than the currently available heuristic.

math.OC↗

arXiv2023

Welfare Maximization Algorithm for Solving Budget-Constrained Multi-Component POMDPs

Partially Observable Markov Decision Processes (POMDPs) provide an efficient way to model real-world sequential decision making processes. Motivated by the problem of maintenance and inspection of a group of infrastructure components with independent dynamics, this paper presents an algorithm to find the optimal policy for a multi-component budget-constrained POMDP. We first introduce a budgeted-POMDP model (b-POMDP) which enables us to find the optimal policy for a POMDP while adhering to budget constraints. Next, we prove that the value function or maximal collected reward for a b-POMDP is a concave function of the budget for the finite horizon case. Our second contribution is an algorithm to calculate the optimal policy for a multi-component budget-constrained POMDP by finding the optimal budget split among the individual component POMDPs. The optimal budget split is posed as a welfare maximization problem and the solution is computed by exploiting the concave nature of the value function. We illustrate the effectiveness of the proposed algorithm by proposing a maintenance and inspection policy for a group of real-world infrastructure components with different deterioration dynamics, inspection and maintenance costs. We show that the proposed algorithm vastly outperforms the policy currently used in practice.

math.OC↗

arXiv2023

Quantitative Resilience of Generalized Integrators

To design critical systems engineers must be able to prove that their system can continue with its mission even after losing control authority over some of its actuators. Such a malfunction results in actuators producing possibly undesirable inputs over which the controller has real-time readings but no control. By definition, a system is resilient if it can still reach a target after a partial loss of control authority. However, after such a malfunction, a resilient system might be significantly slower to reach a target compared to its initial capabilities. To quantify this loss of performance we introduce the notion of quantitative resilience as the maximal ratio of the minimal times required to reach any target for the initial and malfunctioning systems. Naive computation of quantitative resilience directly from the definition is a complex task as it requires solving four nested, possibly nonlinear, optimization problems. The main technical contribution of this work is to provide an efficient method to compute quantitative resilience of control systems with multiple integrators and nonsymmetric input sets. Relying on control theory and on two novel geometric results we reduce the computation of quantitative resilience to a linear optimization problem. We illustrate our method on two numerical examples: a trajectory controller for low-thrust spacecrafts and a UAV with eight propellers.

eess.SY↗

arXiv2023

Optimal Routing of Modular Agents on a Graph

Motivated by an emerging framework of Autonomous Modular Vehicles, we consider the abstract problem of optimally routing two modules, i.e., vehicles that can attach to or detach from each other in motion on a graph. The modules' objective is to reach a preset set of nodes while incurring minimum resource costs. We assume that the resource cost incurred by an agent formed by joining two modules is the same as that of a single module. Such a cost formulation simplistically models the benefits of joining two modules, such as passenger redistribution between the modules, less traffic congestion, and higher fuel efficiency. To find an optimal plan, we propose a heuristic algorithm that uses the notion of graph centrality to determine when and where to join the modules. Additionally, we use the nearest neighbor approach to estimate the cost routing for joined or separated modules. Based on this estimated cost, the algorithm determines the subsequent nodes for both modules. The proposed algorithm is polynomial time: the worst-case number of calculations scale as the eighth power of the number of the total nodes in the graph. To validate its benefits, we simulate the proposed algorithm on a large number of pseudo-random graphs, motivated by real transportation scenario where it performs better than the most relevant benchmark, an adapted nearest neighbor algorithm for two separate agents, more than 85 percent of the time.

math.OC↗

arXiv2023

Resilience of Linear Systems to Partial Loss of Control Authority

After a loss of control authority over thrusters of the Nauka module, the International Space Station lost attitude control for 45 minutes with potentially disastrous consequences. Motivated by this scenario, we investigate the continued capability of control systems to perform their task despite partial loss of authority over their actuators. We say that a system is resilient to such a malfunction if for any undesirable inputs and any target state there exists an admissible control driving the state to the target. Building on controllability conditions and differential games theory, we establish a necessary and sufficient condition for the resilience of linear systems. As their task might be time-constrained, ensuring completion alone is not sufficient. We also want to estimate how much slower the malfunctioning system is compared to its nominal performance. Relying on Lyapunov theory we derive analytical bounds on the reach times of the nominal and malfunctioning systems in order to quantify their resilience. We illustrate our work on the ADMIRE fighter jet model and on a temperature control system.

eess.SY↗