ACT: Automated CPS Testing for Open-Source Robotic Platforms Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04 Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04

Aditya A. Krishnan, Donghoon Kim, Hokeun Kim · 2026

Read ACT: Automated CPS Testing for Open-Source Robotic Platforms Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04 Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04 by Aditya A. Krishnan, Donghoon Kim, Hokeun Kim (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-11708.txt.

Introduction

Krishnan, Donghoon Kim, and Hokeun Kim present ACT, Automated CPS Testing, to help open-source teams check robot-linked software lacking thorough testing. With many volunteers writing separate CPS modules, tests on robotic platforms are often missing, so serious faults stay hidden. ACT fills that gap with automated, steady checks of community code on its robots, linked to open-source infrastructure such as GitHub. They built an early version and ran a case study on a public CPS with an educational robot platform, showing what it can do.

Catalog record

CREATORAditya A. Krishnan, Donghoon Kim, Hokeun Kim
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE21742
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.11708

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

ACT: Automated CPS Testing for Open-Source Robotic Platforms Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04 Conference: 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026); April 13–14, 2026; Rio de Janeiro, Brazil 7th ACM/IEEE International Conference on Automation of Software Test (AST 2026) (AST ’26), April 13–14, 2026, Rio de Janeiro, Brazil DOI: 10.1145/3793654.3793741 ISBN: 979-8-4007-2476-3/2026/04 Aditya A. Krishnan, Donghoon Kim, Hokeun Kim Abstract . Open-source software for cyber-physical systems (CPS) often lacks robust testing involving robotic platforms, resulting in critical errors that remain undetected. This is especially challenging when multiple modules of CPS software are developed by various open-source contributors. To address this gap, we propose A utomated C PS T esting (ACT) that performs automated, continuous testing of open-source software with its robotic platforms, integrated with the open-source infrastructure such as GitHub. We implement an ACT prototype and conduct a case study on an open-source CPS with an educational robotic platform to demonstrate its capabilities. † † cc-license: by † † cc-license: by 1. Introduction Open-source software (OSS) for cyber-physical systems (CPS) and robotics, such as Autoware ( 3 ) or Robotic Operating System 2 (ROS2) ( 17 ) have been widely used for various CPS, including autonomous vehicles, industrial robots, and embedded control systems. Frequent interactions with the physical world and humans make the safety and correctness of CPS critical; thus, rigorous testing for open-source CPS is sorely needed , contributed by community developers who can unintentionally introduce bugs or errors. However, the current CPS OSS projects face severe challenges. For example, the majority of testing infrastructure of CPS OSS relies on simulation ( 4 ) or abstraction ( 18 ) of the underlying hardware; thus, it does not guarantee correctness of the software when deployed on CPS hardware ( 1 ) . In addition, general challenges in the testing of OSS, such as limited automation ( 16 ) or insufficient coverage ( 13 ) , become more prominent for building CPS testbeds ( 22 ) . Another important challenge in open-source CPS testing is the high cost and extensive latency of continuous integration and continuous deployment (CI/CD) ( 30 ; 27 ) of CPS OSS. All in all, these challenges render critical issues and problems in the open-source CPS software undetected and shipped to real-world deployment. To address this problem, this paper proposes a continuous, cost-effective, A utomated C PS T esting (ACT) platform. The proposed ACT platform can provide low-cost, low-latency testing of robotic components widely used by open-source CPS software, including testing of status lights, inertial measurement units (IMU), general physical sensors, displays, and actuators such as motors. This paper’s contributions are threefold: (i) proposal of the ACT approach for open-source CPS software with OSS infrastructure (e.g., GitHub) using a local self-hosted runner ( 10 ) integrated with robotic platforms, (ii) in-depth case study of ACT implementation with real-world CPS OSS called Lingua Franca (LF) ( 23 ) with test cases from educational CPS lab exercises ( 11 ) based on LF using a commercial-off-the-shelf robotic hardware, Pololu 3pi+ 2040 Robot ( 8 ) , and (iii) analysis and diagnosis results demonstrating capabilities and viability of ACT with regards to important topics on testing CPS OSS such as flakiness, testing latency, and stability. 2. Related Work Testing infrastructure for OSS, such as CI/CD for CPS, has been a bottleneck due to multiple reasons, as identified by a survey on 20 OSS ( 29 ) . One notable bottleneck is resource availability and specific build environments, due to multiple interacting systems that require different software versions in complex CPS. Another study on industrial CPS projects ( 30 ) has shown that unstable testing environments and limited resources are significant challenges in most organizations. A step towards creating a framework that can be used for testing in CPS would be to have a deterministic and stable build environment paired with improved automated testing that does not hinder already established CI/CD pipelines. There has been a rich literature on improving CPS OSS. A testbed by Shrivastava et al. ( 26 ) focuses on testing the timing behavior via various methods using specific testbed setups. Another study ( 21 ) on various robotic testbeds shows the plethora of robots and techniques that are used to test algorithms and robot behaviors. Online platforms for testing robotic algorithms with physical robots are designed to verify the algorithms that were developed either for educational or prototyping purposes, like the Robot Programming Network (RPN) ( 6 ) and RoboRacer (formerly known as F1TENTH ( 14 ) ). Similarly, this paper also uses educational labs for CPS and embedded systems software as a case study for the proposed approach. Sadri-Moshkenani et al. ( 25 ) present a comprehensive survey of test-case generation, selection, and prioritization for CPS, characterizing the properties of state-of-the-art techniques. They also delineate unresolved challenges and outline open research directions across CPS testing. Noticeable among these is the difficulty of designing reliable test oracles for nondeterministic ( 28 ) , continuous behaviors ( 5 ) , which often require tolerance bands or acceptable sets rather than single expected values, and the inadequacy of traditional code coverage, which calls for signal and feature-oriented adequacy criteria tailored to continuous, real-time dynamics. 3. Proposed Automated CPS Testing Approach Figure 1. Proposed ACT platform for automated testing CPS OSS with robotic hardware; (I) ACT runtime with software and a self-hosted runner, (II) ACT hardware for testing robotic platform, (a) local machine for the self-hosted runner and ACT software, and (b) overall flow inside ACT software. Figure 1 illustrates the proposed approach, ACT , as a robust automated testing framework for CPS with physical robotic hardware. Our framework consists of two main components; (i) ACT runtime (Figure 1 -I) which includes of the GitHub self-hosted runner ( 10 ) and the software that runs the testing and (ii) ACT hardware (Figure 1 -II) that controls the hardware testbed responsible controlling the hardware-based testing mechanisms such as the pan and tilt platform for testing inertial measurement unit (IMU) sensors. ACT runtime (Figure 1 -I) runs on a local machine (Figure 1 a) for testing and verification of OSS. The local machine can be an embedded board (e.g., Raspberry Pi) or a server connected to networks and ACT hardware (Figure 1 -II). Firstly, our GitHub action ( 9 ) automatically triggers the self-hosted runner when testing is needed, e.g., before merging certain pull requests (PRs). ACT provides triggering conditions (e.g., merge requests and specific labels) for specific test cases for each peripheral, and these conditions can be configured by the users. We also provide fine control over which test case to trigger to avoid constant triggering of the test cases. The workflow will take care of the cloning and compiling of the programs, and the Python software will monitor the test cases and report the values along with the graphs generated. We make our ACT runtime open-sourced for follow-up research. 1 1 1 Available online: https://github.com/lf-lang/act-lf-testbed ACT software’s internal

Read the complete text