Searcharxiv⌕ Search

arXiv subjects

Bruce Maxim

Publications and source records attributed to Bruce Maxim.

3 recordsLinked to original sources

Mining Unity VR Repositories for Refactoring Patterns and Technical Debt

Maintaining virtual reality (VR) applications poses challenges tied to real-time performance, hardware dependencies, and complex interactions. Yet little is known about how developers refactor these applications, which kinds of technical debt they address, or how well automated tools capture VR-specific changes. We mined 1,755 candidate refactoring commits from 176 open-source Unity VR repositories and manually examined a sample of 303 commits, of which 270 contained at least one refactoring. We also extended RefactoringMiner to create VRRefMiner, a tool that detects conventional refactorings in C# scripts used by Unity applications. Our analysis found both conventional and VR-specific refactorings. VR-specific changes addressed concerns such as interaction logic, performance, and Unity-specific artifacts. Code debt was the most prevalent category overall, while interaction debt and UI/UX debt were the main categories addressed by VR-specific refactorings. Across the repositories, VRRefMiner detected 15,549 operations spanning 68 refactoring types, with structural moves and extractions accounting for 31.0% of detections. It detected conventional refactorings in 216 of 246 commits labeled as general refactoring. In the 24 commits labeled as VR-specific, the tool could detect conventional changes but identified no VR-specific refactoring categories. These results point to a need for detection techniques that account for both source code and the Unity artifacts involved in VR maintenance.

cs.SE↗

Teach-to-Crash: A Closed-Loop Student-Teacher LLM Framework for Collision-Inducing Test Scenario Generation

Validating Autonomous Driving Systems (ADS) in simulation requires testing architectures that can discover rare, safety-critical failures while generating scenarios that are executable, diverse, and useful for downstream failure analysis. We introduce Teach-to-Crash, a closed-loop testing framework that combines a constrained ego-centric scenario representation, stagnation-aware search control, and a dual-LLM architecture for adaptive failure discovery. A high-reasoning Teacher LLM acts as an adaptive search controller, while a low-reasoning Student LLM emits simulator-executable scenarios in a strict JSON schema. The Teacher intervenes only when rolling collision rate and time-to-collision metrics stagnate, providing strategic guidance to redirect the search. In a CARLA case study with two experimental setups that vary the ego vehicle's speed policy, Teach-to-Crash achieves the highest Collision Hit Rate (90.79%), the shortest mean Time-to-Collision (18.31 s), and a competitive Collision Discovery Rate (136.21). PAFOT attains a higher mean CDR (179.44), but with substantially larger variance. Teach-to-Crash also yields the highest diversity (0.547) and, averaged across both setups on the CARLA Traffic Manager controller, the highest avoidability-based usefulness proxy (60.04%) among the compared methods. These results, within the evaluated CARLA scope, provide evidence that closed-loop dual-LLM reasoning can steer adversarial simulation-based testing over a constrained executable program space, generating failures that are frequent, structurally diverse, and assessed as more frequently avoidable.

cs.SE↗

Analyzing Performance Issues of Virtual Reality Applications

Extended Reality (XR) includes Virtual Reality (VR), Augmented Reality (AR) and Mixed Reality (MR). XR is an emerging technology that simulates a realistic environment for users. XR techniques have provided revolutionary user experiences in various application scenarios (e.g., training, education, product/architecture design, gaming, remote conference/tour, etc.). Due to the high computational cost of rendering real-time animation in limited-resource devices and constant interaction with user activity, XR applications often face performance bottlenecks, and these bottlenecks create a negative impact on the user experience of XR software. Thus, performance optimization plays an essential role in many industry-standard XR applications. Even though identifying performance bottlenecks in traditional software (e.g., desktop applications) is a widely explored topic, those approaches cannot be directly applied within XR software due to the different nature of XR applications. Moreover, XR applications developed in different frameworks such as Unity and Unreal Engine show different performance bottleneck patterns and thus, bottleneck patterns of Unity projects can't be applied for Unreal Engine (UE)-based XR projects. To fill the knowledge gap for XR performance optimizations of Unreal Engine-based XR projects, we present the first empirical study on performance optimizations from seven UE XR projects, 78 UE XR discussion issues and three sources of UE documentation. Our analysis identified 14 types of performance bugs, including 12 types of bugs related to UE settings issues and two types of CPP source code-related issues. To further assist developers in detecting performance bugs based on the identified bug patterns, we also developed a static analyzer, UEPerfAnalyzer, that can detect performance bugs in both configuration files and source code.

cs.SE↗