arXiv subjects

Joonwon Choi

Publications and source records attributed to Joonwon Choi.

4 recordsLinked to original sources

arXiv2026

A Barrier-Free Synchronization Algorithm for Multi-Engine AI Accelerators

Multi-engine AI accelerators such as AWS Trainium comprise specialized compute engines that execute in parallel, and the compiler must synchronize the data dependencies between them. For straight-line code this is simple: each dependency reduces to waiting for a threshold count of instruction completions, which the compiler computes statically. Loops admit no such static threshold; a simple solution inserts all-engine barriers at iteration boundaries, resetting synchronization state so each loop body can be treated as straight-line, at the cost of parallelism. We present a barrier-free synchronization algorithm that instead enforces each dependency precisely across structured control flow with arbitrarily nested, dynamically bounded loops. The key idea is to compute dynamic thresholds at runtime from tracked loop iteration counts. We implemented it as a compiler backend pass at the AWS Neuron ISA level. On a suite of ML kernels, it reduces latency 10-45% relative to the barrier-based baseline, achieves a 3.3x speedup on a synchronization-bound microbenchmark, and often matches or exceeds hand-tuned manual allocation. Issuing a consumer too early violates its dependency, while issuing too late unnecessarily stalls execution. We formally characterize the minimum synchronization required for correctness and verify in the Lean proof assistant, via bisimulation, that our algorithm meets this criterion.

cs.PL↗

arXiv2026

Reachability-based Temporal Logic Verification for Reliable LLM-guided Human-Autonomy Teaming

We propose a reachability-based framework for reliable LLM-guided human-autonomy teaming (HAT) using signal temporal logic (STL). In the proposed framework, LLM is leveraged as a translator that transfers natural language commands given by a human operator into corresponding STL specifications or vice versa. An STL feasibility filter (SFF) is proposed to check the feasibility of the generated STL. The SFF first decomposes the complex and nested LLM translation into a set of simpler subformulas for parallelization and informative feedback generation. The reachability analysis method is then applied to verify if each subformula is feasible for a target dynamical system: if feasible, perform mission planning, otherwise, reject it. The proposed SFF can identify infeasible subformulas, more than simply providing the boolean verification results for the whole STL, thereby facilitating the feedback generation of LLM to request modification of the command to the human. Consequently, the proposed framework can allow more reliable HAT by enabling safe and informative communication between the human operator and the autonomous agent. Our experiments demonstrate that the proposed framework can successfully filter out infeasible subformulas and generate informative feedback based on such information.

eess.SY↗

arXiv2026

Signal Temporal Logic Verification and Synthesis Using Deep Reachability Analysis and Layered Control Architecture

We propose a signal temporal logic (STL)-based framework that rigorously verifies the feasibility of a mission described in STL and synthesizes control to safely execute it. The proposed framework ensures safe and reliable operation through two phases. First, the proposed framework assesses the feasibility of STL by computing a backward reachable tube (BRT), which captures all states that can satisfy the given STL, regardless of the initial state. The proposed framework accommodates the multiple reach-avoid (MRA) problem to address more general STL specifications and leverages a deep neural network to alleviate the computation burden for reachability analysis, reducing the computation time by about 1000 times compared to a baseline method. We further propose a layered planning and control architecture that combines mixed-integer linear programming (MILP) for global planning with model predictive control (MPC) as a local controller for the verified STL. Consequently, the proposed framework can robustly handle unexpected behavior of obstacles that are not described in the environment information or STL, thereby providing reliable mission performance. Our numerical simulations demonstrate that the proposed framework can successfully compute BRT for a given STL and perform the mission.

eess.SY↗

arXiv2022

State Prediction of Human-in-the-Loop Multi-rotor System with Stochastic Human Behavior Model

Reachability analysis is a widely used method to analyze the safety of a Human-in-the-Loop Cyber Physical System (HiLCPS). This strategy allows the HiLCPS to respond against an imminent threat in advance by predicting reachable states of the system. However, it could lead to an unnecessarily conservative reachable set if the prediction only relies on the system dynamics without explicitly considering human behavior, and thus the risk might be overestimated. To reduce the conservativeness of the reachability analysis, we present a state prediction method which takes into account a stochastic human behavior model represented as a Gaussian Mixture Model (GMM). In this paper, we focus on the multi-rotor in a near-collision situation. The stochastic human behavior model is trained using experimental data to represent human operators' evasive maneuver. Then, we can retrieve a human control input probability distribution from the trained stochastic human behavior model using the Gaussian Mixture Regression (GMR). The proposed algorithm predicts the probability distribution of the multi-rotor's future state based on the given dynamics and the retrieved human control input probability distribution. Besides, the proposed state prediction method considers the uncertainty of the initial state modeled as a GMM, which yields more robust performance. Human subject experiment results are provided to demonstrate the effectiveness of the proposed algorithm.

eess.SY↗