Large language models (LLMs) have become essential tools in computer science, especially for tasks involving code understanding and generation. However, existing work does not address the unique challenges presented by the scale and complexity of code written for government applications. In particular, government enterprise software is often written in legacy languages like MUMPS or assembly language code (ALC) and the overall token lengths of these systems exceed the context window size for current commercially available LLMs. Additionally, LLMs are primarily trained on modern software languages and have undergone limited testing with legacy languages, making their ability to understand legacy languages unknown and, hence, an area for empirical study. This paper examines the application of LLMs in the modernization of legacy government code written in ALC and MUMPS, addressing the challenges of input limitations. We investigate various code-chunking methods to optimize the generation of summary module comments for legacy code files, evaluating the impact of code-chunking methods on the quality of documentation produced by different LLMs, including GPT-4o, Claude 3 Sonnet, Mixtral, and Llama 3. Our results indicate that LLMs can select partition points closely aligned with human expert partitioning. We also find that chunking approaches have significant impact on downstream tasks such as documentation generation. LLM-created partitions produce comments that are up to 20

Large language models (LLMs) have been increasingly integrated into software engineering and maintenance tasks due to their high performance with software engineering tasks and robust understanding of modern programming languages. However, the ability of LLMs to comprehend code written with legacy languages remains a research gap challenged by real-world legacy systems lacking or containing inaccurate documentation that may impact LLM comprehension. To assess LLM comprehension of legacy languages, there is a need for objective LLM evaluation. In order to objectively measure LLM comprehension of legacy languages, we need an efficient, quantitative evaluation method. We leverage multiple-choice question answering (MCQA), an emerging LLM evaluation methodology, to evaluate LLM comprehension of legacy code and the impact of comment prevalence and inaccurate comments. In this work, we present preliminary findings on the impact of documentation on LLM comprehension of legacy code and outline strategic objectives for future work.

Large Language Models (LLMs) have demonstrated impressive abilities in recent years with regards to code generation and understanding. However, little work has investigated how documentation and other code properties affect an LLM's ability to understand and generate code or documentation. We present an empirical analysis of how underlying properties of code or documentation can affect an LLM's capabilities. We show that providing an LLM with "incorrect" documentation can greatly hinder code understanding, while incomplete or missing documentation does not seem to significantly affect an LLM's ability to understand code.

Legacy software systems, written in outdated languages like MUMPS and mainframe assembly, pose challenges in efficiency, maintenance, staffing, and security. While LLMs offer promise for modernizing these systems, their ability to understand legacy languages is largely unknown. This paper investigates the utilization of LLMs to generate documentation for legacy code using two datasets: an electronic health records (EHR) system in MUMPS and open-source applications in IBM mainframe Assembly Language Code (ALC). We propose a prompting strategy for generating line-wise code comments and a rubric to evaluate their completeness, readability, usefulness, and hallucination. Our study assesses the correlation between human evaluations and automated metrics, such as code complexity and reference-based metrics. We find that LLM-generated comments for MUMPS and ALC are generally hallucination-free, complete, readable, and useful compared to ground-truth comments, though ALC poses challenges. However, no automated metrics strongly correlate with comment quality to predict or measure LLM performance. Our findings highlight the limitations of current automated measures and the need for better evaluation metrics for LLM-generated documentation in legacy systems.

In most modern cities, traffic congestion is one of the most salient societal challenges. Past research has shown that inserting a limited number of autonomous vehicles (AVs) within the traffic flow, with driving policies learned specifically for the purpose of reducing congestion, can significantly improve traffic conditions. However, to date, these AV policies have generally been evaluated under the same limited conditions under which they were trained. On the other hand, to be considered for practical deployment, they must be robust to a wide variety of traffic conditions. This article establishes for the first time that a multiagent driving policy can be trained in such a way that it generalizes to different traffic flows, AV penetration, and road geometries, including on multilane roads. Inspired by our successful results in a high-fidelity microsimulation, this article further contributes a novel extension of the well-known cell transmission model (CTM) that, unlike the past CTMs, is suitable for modeling congestion in traffic networks, and is thus suitable for studying congestion reduction policies such as those considered in this article.

Productive and efficient human-robot teaming is a highly desirable ability in service robots, yet there is a fundamental trade-off that a robot needs to consider in such tasks. On the one hand, gaining information from communication with teammates can help individual planning. On the other hand, such communication comes at the cost of distracting teammates from efficiently completing their goals, which can also harm the overall team performance. In this study, we quantify the cost of interruptions in terms of degradation of human task performance, as a robot interrupts its teammate to gain information about their task. Interruptions are varied in timing, content, and proximity. The results show that people find the interrupting robot significantly less helpful. However, the human teammate's performance in a secondary task deteriorates only slightly when interrupted. These results imply that while interruptions can objectively have a low cost, an uninformed implementation can cause these interruptions to be perceived as distracting. These research outcomes can be leveraged in numerous applications where collaborative robots must be aware of the costs and gains of interruptive communication, including logistics and service robots.

Ad hoc teamwork is the research problem of designing agents that can collaborate with new teammates without prior coordination. This survey makes a two-fold contribution: First, it provides a structured description of the different facets of the ad hoc teamwork problem. Second, it discusses the progress that has been made in the field so far, and identifies the immediate and long-term open problems that need to be addressed in ad hoc teamwork.

. Ad-hoc teamwork is defined as the task of collaborating with teammates without pre-coordination. When the ad hoc agent is a robot that needs to collaborate with people, it cannot assume that its team-mates will behave optimally or legibly. By providing a means to learn human policies in ad-hoc teamwork, this work will help create robots that can adapt to a new human agent and work together to achieve a common goal. We focus on a simple, yet powerful model for representing agents using the concept of bounded rationality. Our preliminary results exemplify how such a model can be used in a domain from the ad-hoc teamwork literature called “the tool fetching domain”.

The UT Austin Villa team, from the University of Texas at Austin, won the 2021 RoboCup 3D Simulation League, winning all 19 games the team played. During the course of the competition the team scored 108 goals while conceding only 5. Additionally the team finished second in the overall RoboCup 3D Simulation League technical challenge by finishing second in both the fat proxy and scientific challenges. This paper details and analyzes the results of the 2021 competition, and also presents a new deep RL learning framework that was presented during the scientific challenge.

A desirable goal for autonomous agents is to be able to coordinate on the fly with previously unknown teammates. Known as "ad hoc teamwork", enabling such a capability has been receiving increasing attention in the research community. One of the central challenges in ad hoc teamwork is quickly recognizing the current plans of other agents and planning accordingly. In this paper, we focus on the scenario in which teammates can communicate with one another, but only at a cost. Thus, they must carefully balance plan recognition based on observations vs. that based on communication. This paper proposes a new metric for evaluating how similar are two policies that a teammate may be following - the Expected Divergence Point (EDP). We then present a novel planning algorithm for ad hoc teamwork, determining which query to ask and planning accordingly. We demonstrate the effectiveness of this algorithm in a range of increasingly general communication in ad hoc teamwork problems.

Traffic congestion is a major challenge in modern urban settings. The industry-wide development of autonomous and automated vehicles (AVs) motivates the question of how can AVs contribute to congestion reduction. Past research has shown that in small scale mixed traffic scenarios with both AVs and human-driven vehicles, a small fraction of AVs executing a controlled multiagent driving policy can mitigate congestion. In this paper, we scale up existing approaches and develop new multiagent driving policies for AVs in scenarios with greater complexity. We start by showing that a congestion metric used by past research is manipulable in open road network scenarios where vehicles dynamically join and leave the road. We then propose using a different metric that is robust to manipulation and reflects open network traffic efficiency. Next, we propose a modular transfer reinforcement learning approach, and use it to scale up a multiagent driving policy to outperform human-like traffic and existing approaches in a simulated realistic scenario, which is an order of magnitude larger than past scenarios (hundreds instead of tens of vehicles). Additionally, our modular transfer learning approach saves up to 80% of the training time in our experiments, by focusing its data collection on key locations in the network. Finally, we show for the first time a distributed multiagent policy that improves congestion over human-driven traffic. The distributed approach is more realistic and practical, as it relies solely on existing sensing and actuation capabilities, and does not require adding new communication infrastructure.

Traffic congestion is a major challenge in modern urban settings. The industry-wide development of autonomous and automated vehicles (AVs) motivates the question of how can AVs contribute to congestion reduction. Past research has shown that in small scale mixed traffic scenarios with both AVs and human-driven vehicles, a small fraction of AVs executing a controlled multiagent driving policy can mitigate congestion. In this paper, we scale up existing approaches and develop new multiagent driving policies for AVs in scenarios with greater complexity. We start by showing that a congestion metric used by past research is manipulable in open road network scenarios where vehicles dynamically join and leave the road. We then propose using a different metric that is robust to manipulation and reflects open network traffic efficiency. Next, we propose a modular transfer learning approach and use it to scale up the multiagent driving policy to a realistic simulated scenario that is an order of magnitude larger than past scenarios (hundreds rather than tens of vehicles). Our experimental study shows that the resulting policy improves traffic efficiency over human-driven traffic in a large open network, where existing approaches fail to do so. Another key advantage of our modular transfer learning approach is that it avoids collecting samples from entire network, which saves up to 80% of training and data collection time in our experiments.

In ad hoc teamwork, multiple agents need to collaborate without having knowledge about their teammates or their plans a priori. A common assumption in this research area is that the agents cannot communicate. However, just as two random people may speak the same language, autonomous teammates may also happen to share a communication protocol. This paper considers how such a shared protocol can be leveraged, introducing a means to reason about Communication in Ad Hoc Teamwork (CAT). The goal of this work is enabling improved ad hoc teamwork by judiciously leveraging the ability of the team to communicate. We situate our study within a novel CAT scenario, involving tasks with multiple steps, where teammates' plans are unveiled over time. In this context, the paper proposes methods to reason about the timing and value of communication and introduces an algorithm for an ad hoc agent to leverage these methods. Finally, we introduces a new multiagent domain, the tool fetching domain, and we study how varying this domain's properties affects the usefulness of communication. Empirical results show the benefits of explicit reasoning about communication content and timing in ad hoc teamwork.

The choice of activation function can have a large effect on the performance of a neural network. While there have been some attempts to hand-engineer novel activation functions, the Rectified Linear Unit (ReLU) remains the most commonly-used in practice. This paper shows that evolutionary algorithms can discover novel activation functions that outperform ReLU. A tree-based search space of candidate activation functions is defined and explored with mutation, crossover, and exhaustive search. Experiments on training wide residual networks on the CIFAR-10 and CIFAR-100 image datasets show that this approach is effective. Replacing ReLU with evolved activation functions results in statistically significant increases in network accuracy. Optimal performance is achieved when evolution is allowed to customize activation functions to a particular task; however, these novel activation functions are shown to generalize, achieving high performance across tasks. Evolutionary optimization of activation functions is therefore a promising new dimension of metalearning in neural networks.

Communication in Ad Hoc Teamwork (CAT) is a research area that investigates how communication can be leveraged by an agent that plans in a distributed, multi-agent collaborative environment, even if that agent does not have knowledge about its teammates or their plans a priori. This paper reports our progress in identifying three factors that can impact the complexity of CAT – environment, teammates, and communication protocol. Following the identification of these components, this paper investigates three extensions from existing work that affect each of these factors respectively – richer environments, complex teammate representations, and complex communication protocols. We present new algorithms to compute when to query under these new configurations, as well as preliminary results of their performance.

The choice of activation function can have a large effect on the final accuracy of a neural network. While there have been some attempts to hand-engineer novel activation functions, there remain few activation functions that are commonly used in practice. We show that evolutionary algorithms can discover novel activation functions that outperform commonlyused functions such as ReLU. We define a tree-based search space of potential activation functions and explore the space with exhaustive search, mutation, and crossover. Experiments on the MNIST and CIFAR-10 image datasets show our approach is effective for both fully-connected and convolutional architectures. Finally, we verify that our evolutionary algorithm outperforms random search.

Kidney exchange has been studied extensively from the perspective of market design, and a significant focus has been on better algorithms for finding chains and cycles to increase the number of possible matches. A more dramatic benefit could come from incorporating compatible pairs into the mechanism, but this possibility has been relatively understudied. In order to incentivize a compatible pair to participate in exchange, they must be offered a higher quality match for the recipient that can be performed without adding extra waiting time. In this paper, we make two main contributions to the study of incorporating compatible pairs in exchanges. First, we leverage the recently proposed Living Donor Kidney Profile Index (LKDPI) to measure match quality, and develop a novel simulator (based on data from a major transplant center) for the joint distribution of compatibility and quality across pairs. This simulator allows us to study the benefits of including compatible pairs under different models and assumptions. Second, we introduce a hybrid online/batch matching model with impatient (compatible) and patient (incompatible) pairs to capture the need for immediacy. We introduce new algorithms for matching in this model, including one based on online primal-dual techniques. Overall, our results indicate great potential in terms of both increased numbers of transplants of incompatible pairs (almost doubling the number transplanted) as well as improved match quality for recipients in compatible pairs (increasing expected graft survival by between 1 and 2 years). The results are also promising for hard-to-match subpopulations, including blood group O recipients.

We consider an adaptation of Axelrod's metanorm model, where a population of agents choose between cooperating and defecting in bilateral interactions. Because punishing incurs an enforcement cost, Axelrod proposes using metanorms, to facilitate the stability of a norm of punishing defectors, where those who do not punish defectors can themselves be punished. We present two approaches to study the social effects of such metanorms when agents can choose their interaction partners: (a) a theoretical study, when agent behaviors are static, showing stable social configurations, under all possible relationships between system parameters representing agent payoffs with or without defection, punishment, and meta- punishment, and (b) an experimental evaluation of emergent social configurations when agents choose behaviors to maximize expected utility. We highlight emergent social configurations, including anarchy, a "police" state with cooperating agents who enforce, and a unique "corrupt police" state where one enforcer penalizes all defectors but defects on others!

Unmanned Aerial Vehicles (UAVs) have been traditionally controlled via remote control or by software, which require skill using the remote or expert programming skills. Our goal is to develop a natural mode of directing a drone's actions, akin to the forms of expression one finds between a person and a pet and hence accessible to almost any person without specialized training or expertise in using electronic gadgets. We build on prior work on analyzing video streams to use the video from the drone's on-board camera to enable gesture-based control. Our approach uses a pre-trained convolutional neural network for pose extraction, Haar cascades to identify regions of interest within the UAV's field of view, and a finite state machine to select the drone's action.