Once-for-All (OFA) training enables a single super-net to generate multiple sub-nets tailored to diverse deployment scenarios, supporting flexible trade-offs among accuracy, robustness, and model-size without retraining. However, as the number of supported sub-nets increases, excessive parameter sharing in the backbone limits representational capacity, leading to degraded calibration and reduced overall performance. To address this, we propose SOLAR (Switchable Output Layer for Accuracy and Robustness in Once-for-All Training), a simple yet effective technique that assigns each sub-net a separate classification head. By decoupling the logit learning process across sub-nets, the Switchable Output Layer (SOL) reduces representational interference and improves optimization, without altering the shared backbone. We evaluate SOLAR on five datasets (SVHN, CIFAR-10, STL-10, CIFAR-100, and TinyImageNet) using four super-net backbones (ResNet-34, WideResNet-16-8, WideResNet-40-2, and MobileNetV2) for two OFA training frameworks (OATS and SNNs). Experiments show that SOLAR outperforms the baseline methods: compared to OATS, it improves accuracy of sub-nets up to 1.26%, 4.71%, 1.67%, and 1.76%, and robustness up to 9.01%, 7.71%, 2.72%, and 1.26% on SVHN, CIFAR-10, STL-10, and CIFAR-100, respectively. Compared to SNNs, it improves TinyImageNet accuracy by up to 2.93%, 2.34%, and 1.35% using ResNet-34, WideResNet-16-8, and MobileNetV2 backbones (with 8 sub-nets), respectively. The code of SOLAR is publicly available at: https://github.com/NAIL-UH/SOLAR and its website can be accessed at https://saktx.github.io/solar.github.io/.

Deploying high-performance object detectors on TinyML platforms poses significant challenges due to tight hardware constraints and the modular complexity of modern detection pipelines. Neural Architecture Search (NAS) offers a path toward automation, but existing methods either restrict optimization to individual modules—sacrificing cross-module synergy—or require global searches that are computationally intractable. We propose ELASTIC (Efficient Once for AlLIterAtiveSearch for ObjecTDetectIon on MiCrocontrollers), a unified, hardware-aware NAS framework that alternates optimization across modules (e.g., backbone, neck, and head) in a cyclic fashion. ELASTIC introduces a novelPopulation Passthroughmechanism in evolutionary search that retains high-quality candidates between search stages, yielding faster convergence, up to an 8% final mAP gain, and eliminates search instability observed without population passthrough. In a controlled comparison, empirical results show ELASTIC achieves +4.75% higher mAP and 2× faster convergence than progressive NAS strategies on SVHN, and delivers a +9.09% mAP improvement on PascalVOC given the same search budget. ELASTIC achieves 72.3% mAP on PascalVOC, outperforming MCUNET by 20.9% and TinyissimoYOLO by 16.3%. When deployed on MAX78000/MAX78002 microcontrollers, ELASTICderived models outperform Analog Devices’ TinySSD baselines, reducing energy by up to 71.6 %, lowering latency by up to 2.4×, and improving mAP by up to 6.99 percentage points across multiple datasets. The experimental videos and codes are available on the project website1.

This paper addresses a distributed leader-follower formation control problem for a group of agents, each using a body-fixed camera with a limited field of view (FOV) for state estimation. The main challenge arises from the need to coordinate the agents' movements with their cameras' FOV to maintain visibility of the leader for accurate and reliable state estimation. To address this challenge, we propose a novel perception-aware distributed leader-follower safe control scheme that incorporates FOV limits as state constraints. A Control Barrier Function (CBF) based quadratic program is employed to ensure the forward invariance of a safety set defined by these constraints. Furthermore, new neural network based and double bounding boxes based estimators, combined with temporal filters, are developed to estimate system states directly from real-time image data, providing consistent performance across various environments. Comparison results in the Gazebo simulator demonstrate the effectiveness and robustness of the proposed framework in two distinct environments.

Reinforcement learning (RL) agents need to explore their environment to learn optimal behaviors and achieve maximum rewards. However, exploration can be risky when training RL directly on real systems, while simulation-based training introduces the tricky issue of the sim-to-real gap. Recent approaches have leveraged safety filters, such as control barrier functions (CBFs), to penalize unsafe actions during RL training. However, the strong safety guarantees of CBFs rely on a precise dynamic model. In practice, uncertainties always exist, including internal disturbances from the errors of dynamics and external disturbances such as wind. In this work, we propose a novel safe RL framework built on a robust CBF, where the discrepancy between the nominal and true dynamic models is quantified through a combination of disturbance observation and residual model learning. We demonstrate our results on the Safety-gym benchmark for Point and Car robots on all tasks where we can outperform state-of-the-art approaches that use only residual model learning or a disturbance observer (DOB). We further validate the efficacy of our framework using a physical F1/10 racing car.Videos: https://sites.google.com/view/res-dob-cbf-rl

Autonomous racing is a challenging problem, as the vehicle needs to operate at the friction or handling limits in order to achieve minimum lap times. Autonomous race cars require highly accurate perception, state estimation, planning, and control. Adding to this complexity is the need to accurately identify vehicle model parameters governing lateral tire slip effects, which can evolve over time due to factors such as tire wear and tear. Current approaches to this problem typically either propose offline model identification methods or rely on initial parameters within a narrow range (typically within 15-20% of the actual values). However, these approaches fall short in accounting for significant changes in tire models that can occur during actual races, particularly when pushing the vehicle to its handling limits. We present a unified framework that not only learns the tire model in real time from collected data but also adapts the model to environmental changes, even when the model parameters exhibit substantial deviations. The friction estimation, obtained as a byproduct from the learning results, facilitates the selection of the optimal racing line from a library for adaptive speed planning. We validate our approach through testing in simulators, encompassing a 1:43 scale race car and a full-size car, and also through experiments with a physical F1/10 autonomous race car.

Delays endanger the safety of autonomous systems functioning in the rapidly changing environments of autonomous driving and high-speed racing. Unfortunately, the consideration of delays is often overlooked during controller design or learning-enabled controller training phases prior to deployment in the physical world. This paper systematically and comprehensively addresses both the computation delay arising from nonlinear optimization for control and other inevitable delays caused by actuators. First, we propose a new filtering approach to adaptively estimate the time-variant computation delay. Second, we model actuation dynamics for steering delay. Third, all the constrained optimization is realized in a robust tube model predictive controller. In terms of application merits, our approach is a novel design for a standalone delay-aware controller; in addition, our approach can also serve as a delay compensator for an existing controller. Video (https://youtu.be/nURl_HTW_Mo) and code (https://github.com/dvij542/Delay-aware-Robust-Tube-MPC) are available.

The protection of Industrial Control Systems (ICS) that are employed in public critical infrastructures is of utmost importance due to catastrophic physical damages cyberattacks may cause. The research community requires testbeds for validation and comparing various intrusion detection algorithms to protect ICS. However, there exist high barriers to entry for research and education in the ICS cybersecurity domain due to expensive hardware, software, and inherent dangers of manipulating real-world systems. To close the gap, built upon recently developed 3D high-fidelity simulators, we further showcase our integrated framework to automatically launch cyberattacks, collect data, train machine learning models, and evaluate for practical chemical and manufacturing processes. On our testbed, we validate our proposed intrusion detection model called Minimal Threshold and Window SVM (MinTWin SVM) that utilizes unsupervised machine learning via a one-class SVM in combination with a sliding window and classification threshold. Results show that MinTWin SVM minimizes false positives and is responsive to physical process anomalies. Furthermore, we incorporate our framework with ICS cybersecurity education by using our dataset in an undergraduate machine learning course where students gain hands-on experience in practicing machine learning theory with a practical ICS dataset. All of our implementations have been open-sourced.

To efficiently generate safe trajectories for an autonomous vehicle in dynamic environments, a layered motion planning method with decoupled path and speed planning is widely used. This paper studies speed planning, which mainly deals with dynamic obstacle avoidance given a planned path. The main challenges lie in the optimization in a non-convex space and the trade-off between safety, comfort, and efficiency. First, this work proposes to conduct a search in second-order derivative space for generating a comfort-optimal reference trajectory. Second, by combining abstraction and refinement, an algorithm is proposed to construct a convex feasible space for optimization. Finally, a piecewise Bézier polynomial optimization approach with trapezoidal corridors is presented, which theoretically guarantees safety and significantly enlarges the solution space compared with the existing rectangular corridors-based approach. We validate the efficiency and effectiveness of the proposed approach in simulations.

A control barrier functions-based quadratic programming (CBF-QP) method has emerged as a controller synthesis tool to assure safety of autonomous systems owing to the appealing safe forward invariant set. However, the provable safety relies on a precisely described dynamic model, which is not always available in practice. Recent works leverage learning to compensate model uncertainty for a CBF controller. However, these approaches based on reinforcement learning or episodic learning are limited to dealing with time-invariant uncertainty. Also, the reinforcement learning approach learns the uncertainty offline, while episodic learning only updates the controller after a batch of data is available by the end of an episode. Instead, we propose a novel tuning extreme learning machine (tELM)-based CBF controller that can compensate time-variant and time-invariant model uncertainty adaptively in an online manner. We validate our approach's effectiveness in a simulation of an Adaptive Cruise Control (ACC) system.

With the advancement of affordable self-driving vehicles using complicated nonlinear optimization but limited computation resources, computation time becomes a matter of concern. Other factors such as actuator dynamics and actuator command processing cost also unavoidably cause delays. In high-speed scenarios, these delays are critical to the safety of a vehicle. Recent works consider these delays individually, but none unifies them all in the context of autonomous driving. Moreover, recent works inappropriately consider computation time as a constant or a large upper bound, which makes the control either less responsive or over-conservative. To deal with all these delays, we present a unified framework by 1) modeling actuation dynamics, 2) using robust tube model predictive control, and 3) using a novel adaptive Kalman filter without assuming a known process model and noise covariance, which makes the controller safe while minimizing conservativeness. On the one hand, our approach can serve as a standalone controller; on the other hand, our approach provides a safety guard for a high-level controller, which assumes no delay. This can be used for compensating the sim-to-real gap when deploying a black-box learning-enabled controller trained in a simplistic environment without considering delays for practical vehicle systems.

This work proposes a novel framework for motion planning using trajectory optimization for autonomous driving. First, a two-phase behavioral policy maker (BPM) is proposed as a high-level decision maker to mimic human-like driving style by avoiding unnecessary tasks and early lane changes. Second, a comprehensive study on iterative adaptive weight tuning functions has been done to limit manual weight tuning in the Constrained Iterative Linear Quadratic Regulator (CILQR) motion planner. Third, a jerk-minimized CILQR is presented to ensure the comfort and safety of passengers by generating smooth trajectories. The simulation results show efficiency, safety, and comfort of generated trajectories.

Learning-enabled components have been widely deployed in autonomous systems. However, due to the weak interpretability and the prohibitively high complexity of large-scale machine learning models such as neural networks, reliability has been a crucial concern for safety-critical autonomous systems. This work proposes an online monitor called Reach-Flow for fault prevention of waypoint-following tasks for self-driving cars. It mainly consists of two components: (a) an online verification tool which conservatively checks the safety of the system behavior in the near future, and (b) a fallback controller which steers the system back to a desired state when the system is potentially unsafe. We implement ReachFlow in a self-driving racing car governed by a reinforcement learning-based controller. We demonstrate the effectiveness by rigorously verifying a safe waypoint-following control and providing a fallback control for an unsafe situation in which a large deviation from the planned path is predicted.

Constrained Iterative Linear Quadratic Regulator (CILQR), a variant of ILQR, has been recently proposed for motion planning problems of autonomous vehicles to deal with constraints such as obstacle avoidance and reference tracking. However, the previous work considers either deterministic trajectories or persistent prediction for target dynamical obstacles. The other drawback is lack of generality - it requires manual weight tuning for different scenarios. In this paper, two significant improvements are achieved. Firstly, a two-stage uncertainty-aware prediction is proposed. The short-term prediction with safety guarantee based on reachability analysis is responsible for dealing with extreme maneuvers conducted by target vehicles. The long-term prediction leveraging an adaptive least square filter preserves the long-term optimality of the planned trajectory since using reachability only for long-term prediction is too pessimistic and makes the planner over-conservative. Secondly, to allow a wider coverage over different scenarios and to avoid tedious parameter tuning case by case, this paper designs a scenario-based analytical function taking the states from the ego vehicle and the target vehicle as input, and carrying weights of a cost function as output. It allows the ego vehicle to execute multiple behaviors (such as lane-keeping and overtaking) under a single planner. We demonstrate safety, effectiveness, and real-time performance of the proposed planner in simulations.

Imitation learning provides a way to automatically construct a controller by mimicking human behavior from data. For safety-critical systems such as autonomous vehicles, it can be problematic to use controllers learned from data because they cannot be guaranteed to be collision-free. Recently, a method has been proposed for learning a multi-mode hybrid automaton cruise controller (MOHA). Besides being accurate, the logical nature of this model makes it suitable for formal verification. In this paper, we demonstrate this capability using the SpaceEx hybrid model checker as follows. We develop an automated tool to translate the automaton model into constraints and equations required by SpaceEx. We then verify that a pure MOHA controller is not collision-free. By adding a safety state based on headway in time, a rule that human drivers should follow anyway, we do obtain a provably safe cruise control. Moreover, the safe controller remains more humanlike than existing cruise controllers.

This paper proposes a novel hybrid model for learning discrete and continuous dynamics of car-following behaviors. Multiple modes representing driving patterns are identified by partitioning the model into groups of states. The model is visualizable and interpretable for car-following behavior recognition, traffic simulation, and human-like cruise control. The experimental results using the next generation simulation datasets demonstrate its superior fitting accuracy over conventional models.

Industrial Control Systems (ICS) such as water and power are critical to any society. Process anomaly detection mechanisms have been proposed to protect such systems to minimize the risk of damage or loss of resources. In this paper, a graphical model-based approach is proposed for profiling normal operational behavior of an operational ICS referred to as SWaT (Secure Water Treatment). Timed automata are learned as a model of regular behaviors shown in sensors signal like fluctuations of water level in tanks. Bayesian networks are learned to discover dependencies between sensors and actuators. The models are used as a one-class classifier for process anomaly detection, recognizing irregular behavioral patterns and dependencies. The detection results can be interpreted and the abnormal sensors or actuators localized due to the interpretability of the graphical models. This approach is applied to a dataset collected from SWaT. Experimental results demonstrate the model's superior performance on both precision and run-time over methods including support vector machine and deep neural networks. The underlying idea is generic and applicable to other industrial control systems such as power and transportation.

We present a novel way to detect infected hosts and identify malware in networks by analyzing network communication statistics with state-of-the-art automata learning algorithms. The automata encode patterns of short-term interactions in known malicious hosts, and are used to obtain small but effective fingerprints of machine behavior. We showcase the effectiveness of our system, named BASTA1(Behavioral Analytics System using Timed Automata), on a public dataset containing Netflow traces of real-world botnet malware. Compared to a deep packet inspection of communication content, Netflows are easy and cheap to collect and analyze, and preserve a greater degree of privacy. Even though the high level of abstraction in Netflow data makes it more difficult to utilize it, BASTA shows very impressive results achieving high accuracy in several settings while returning few false positives. It is also capable of detecting infections of previously unseen malware.