ACT: Automated CPS Testing for Open-Source Robotic Platforms

Aditya A. Krishnan; Donghoon Kim; Hokeun Kim

cs.ROApr 13, 2026

ACT: Automated CPS Testing for Open-Source Robotic Platforms

Authors: Aditya A. Krishnan, Donghoon Kim, Hokeun Kim

Organizations: Arizona State University · Arkansas State University

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 Automated CPS Testing (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.

Figures & tables

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 2. Example LED status indicator blink detection using a color filter; (a) LED off: no active pixels detected, other LEDs not shown due to filtering. (b) LED on: the mask highlights only the LED, filtered by its HSV (hue, saturation and value).

Figure 3. Deviation (in absolute percentage) between actual and measured LED blink periods over elapsed testing time from our case study for different actual blink periods.

Figure 4. The IMU readings on the robot’s display (left) and those measured by ACT’s camera (right).

Mode	Measured Pitch	Actual Pitch	Measured Roll	Actual Roll	Check
Normal	24.14	25.00	0.20	0.00	Passed
Normal	0.0	0.00	26.07	25.00	Passed
Normal	10.73	10.00	0.36	0.00	Passed
Faulty	217.81	20.00	2.96	0.0	Failed
Faulty	26.56	0.0	208.10	20.00	Failed
Faulty	44.05	0.0	-400.58	-20.00	Failed

Table 1. Measured IMU sensor data and actual angles under normal/faulty conditions (values are in degrees).

Figure 5. Setup for the bump experiment; a servo motor controlled by ACT software rotates (a) a pivoting block to mechanically press the robot’s (b1) right or (b2) left bump sensors on (c) Pololu 3pi+ 2040 Robot, generating bump events.

Fault Classes	Types of Faults	Tested Sensors and Measured (Reported) Results			Display Status	Diagnosis
Fault Classes	Types of Faults	Left Bump	Right Bump	Both Bumps	Display Status	Diagnosis
Hardware (HW)	Left bump sensor fault	No response to bump	Responds to right bump	Responds only to right-bump sensor	Partial or missing data	Left bump sensor hardware fault
	Right bump sensor fault	Responds to left bump	No response to bump	Responds only to left-bump sensor	Partial or missing data	Right bump sensor hardware fault
	Display fault	Normal bump detection	Normal bump detection	Normal bump detection	Incorrect or missing output	Display hardware fault
	Either the display or both of the bump sensors are faulty	Partial or no output	Partial or no output	Partial or no output	No display output	Both sensors are faulty
Software (SW)	Left sensor incorrectly mapped to right	Reports right-bump event	Reports right-bump event	Reports right-bump event	Normal	Left bump sensor mapping error

Table 2. Experimental results and diagnostics for bump sensors and display under injected faults.

Figure 6. Wheel-motion tracking for RPM estimation using a video feed with the (a) wheel stationary (RPM $\approx 0$ ) and the (b) wheel rotating at RPM $\approx 79$ .

Automated Software Testing

Automated software testing uses tools and algorithms to generate, run, and evaluate tests that find defects and prevent regressions.

See all 135 papers

Environment-Grounded Multi-Agent Workflow for Autonomous Penetration Testing
Mar 25, 2026Michael Somma, Markus Großpointner, Paul Zabalegui +2Automated Penetration Testing LLMs for Cybersecurity
JOANNEUM RESEARCH Forschungsgesellschaft mbH, DIGITAL – Institute for Digital Technologies, Steyrergasse 17, 8010 Graz, Austria · TU Graz, Institute for Technical Informatics, Inffeldgasse 16/I, 8010 Graz, Austria · Alias Robotics S.L., Calle Venta de la Estrella, 6 Pab. 130, 01006 Vitoria-Gasteiz, Álava, Spain +1
Design and Characterization of a Unified Continuum Robot Actuation Platform
Oct 14, 2026Yunti Xu, Aedan Mangan, Miheer Potdar +2Modular Robotics Continuum Robot Design
Department of Mechanical and Aerospace Engineering, University of California, San Diego, CA, 92093, USA.
CSAR: Containerized System Architecture for Robotics
Jun 29, 2026Ambrosio-Cestero, Gregorio, Galindo Andrades +5Edge Computing Robotics
Machine Perception and Intelligent Robotics group (MAPIR), Dept. of System Engineering and Automation, Málaga Institute for Mechatronics Engineering and Cyber-Physical Systems (IMECH.UMA), University of Málaga, Blvr. Louis Pasteur 35, 29071 Málaga, Spain.

Mar 25, 2026J/K move · Enter open · S save

Environment-Grounded Multi-Agent Workflow for Autonomous Penetration Testing

Michael Somma, Markus Großpointner, Paul Zabalegui, Eppu Heilimo+1

JOANNEUM RESEARCH Forschungsgesellschaft mbH, DIGITAL – Institute for Digital Technologies, Steyrergasse 17, 8010 Graz, Austria · TU Graz, Institute for Technical Informatics, Inffeldgasse 16/I, 8010 Graz, Austria · Alias Robotics S.L., Calle Venta de la Estrella, 6 Pab. 130, 01006 Vitoria-Gasteiz, Álava, Spain+1

Automated Penetration Testing LLMs for Cybersecurity Multi-Agent Systems

Open paper AskPDF arXiv

The increasing complexity and interconnectivity of digital infrastructures make scalable and reliable security assessment methods essential. Robotic systems represent a particularly important class of operational technology, as modern robots are highly networked cyber-physical systems deployed in domains such as industrial automation, logistics, and autonomous services. This paper explores the use of large language models for automated penetration testing in robotic environments. We propose an environment-grounded multi-agent architecture tailored to Robotics-based systems. The approach dynamically constructs a shared graph-based memory during execution that captures the observable system state, including network topology, communication channels, vulnerabilities, and attempted exploits. This enables structured automation while maintaining traceability and effective context management throughout the testing process. Evaluated across multiple iterations within a specialized robotics Capture-the-Flag scenario (ROS/ROS2), the system demonstrated high reliability, successfully completing the challenge in 100% of test runs (n=5). This performance significantly exceeds literature benchmarks while maintaining the traceability and human oversight required by frameworks like the EU AI Act.

Abstract

Automated Software Testing

Explore similar work

Environment-Grounded Multi-Agent Workflow for Autonomous Penetration Testing

Design and Characterization of a Unified Continuum Robot Actuation Platform

CSAR: Containerized System Architecture for Robotics