Radeis Extension for NVIDIA Isaac Sim is designed for pre-deployment robotic cyber-safety testing. Supply: VicOne
As issues about cybersecurity improve, robotics builders are getting instruments to check it in simulation. VicOne Inc. yesterday launched the free VicOne Radeis Extension for NVIDIA Isaac Sim, which it mentioned allows builders to check how chosen assault eventualities may have an effect on the habits of their very own robotic fashions.
āRobotics firms are shifting from proving particular person use instances to deploying robots at scale in real-world environments. As that shift accelerates, the business should validate not solely whether or not robots operate as designed, however [also] whether or not they stay dependable when the digital techniques they rely on are compromised,ā acknowledged Max Cheng, CEO of VicOne.
āCyber-safety validation should turn into a normal a part of robotics growth,” he added. “By turning DEF CON analysis into repeatable simulation eventualities, we’re serving to builders deal with these dangers earlier and construct extra resilient bodily AI at scale.ā
Based in 2022, VicOne has constructed its cybersecurity for bodily AI on its automotive experience. The Pattern Micro subsidiary mentioned its software program and companies can defend the digital techniques that form how automobiles and robots see, resolve, and act.
Tokyo-based VicOne helps OEMs, Tier 1 suppliers, robotic makers, and operators establish cyber dangers early, assess their potential security affect, and defend techniques in operation. The company mentioned it has uncovered greater than 180 zero-days throughout automotive and robotics, added over 100 million risk intelligence alertsĀ month-to-month, and filed 30 U.S. patent purposes for AI-defined car and bodily AI safety.
VicOne learns from DEF CON findings
Robotic makers make investments closely in practical testing, security engineering, and simulation, however cyber-induced circumstances can nonetheless fall outdoors typical check plans, famous VicOne. At DEF CON 34’s Robotic Hacking Neighborhood, the corporate carried out its Physical AI Safety Stress TestĀ CTF (seize the flag) train.
Safety researchers examined how digital compromises may alter robotic habits underneath managed circumstances. The problem examined assault paths throughout AI fashions, notion, cloud and API companies, robotics communications, and embedded techniques.
The DEF CON actions included the Reachy Mini and LeKiwi robotics platforms. They supplied researchers with hands-on techniques for exploring how cybersecurity vulnerabilities couldaffect bodily robotic habits.
VicOne studied unsuccessful makes an attempt and newly noticed assault assumptions to refine eventualities involving manipulated visible inputs, hijacked instructions, poisoned AI insurance policies, and exploited related interfaces. By testing these circumstances on the system stage, groups can decide whether or not vulnerabilities in particular person elements may finally alter the habits of the whole robotic, it mentioned.
At DEF CON 34, 16 groups recorded 25 profitable solves, famous VicOne. Almost each core AI and robotics cyber-safety problem was solved not less than as soon as, with just one superior embedded techniques problem involving cryptography and firmware vulnerabilities remaining unsolved.
The corporate mentioned outcomes confirmed that behavior-changing dangers can emerge throughout a number of layers of the robotic stack, reinforcing the necessity to consider cybersecurity on the system stage slightly than assessing vulnerabilities solely in isolation.
Safety researchers, robotics consultants, and hackers gathered on the Robotic Hacking Neighborhood at DEF CON 34 to discover how cyber threats can have an effect on AI-powered robots. Supply: VicOne
Extension applies robotic hacking analysis to repeatable testing
Each assault try can contribute to stronger safety, asserted VicOne. A profitable try could reveal a path from digital compromise to modified robotic habits. An unsuccessful try can nonetheless expose attacker assumptions, unexplored management paths, or safeguards that require additional validation.
VicOne mentioned it interprets these findings into structured eventualities that builders can repeat as robotic software program, sensors, AI fashions, communications, and management architectures evolve. Utilizing the free extension, builders can:
- Carry their robotic mannequin and supposed job into NVIDIA Isaac Sim
- Apply curated eventualities knowledgeable by VicOneās bodily AI safety analysis and DEF CON findings
- Validate whether or not a vision-language mannequin (VLM) stays resilient towards adversarial visible prompts, malicious inputs, and sensor manipulation makes an attempt that might result in incorrect interpretations or unsafe decision-making
- Repeat assessments as fashions, software program, sensors, communications, and management paths change
NVIDIA Isaac Sim gives a bodily based mostly surroundings for designing, simulating, testing, and validating AI-driven robots, claimed the corporate. The VicOne Radeis Extension provides cyber-safety eventualities and behavioral-impact remark to present simulation workflows, serving to groups look at security-related edge instances earlier than shifting into bodily testing or deployment.
Builders can then observe whether or not the robotic deviates from its supposed job, working boundaries, or anticipated response. The Radeis Extension permits builders to make use of their very own robotic fashions, supposed duties, and working assumptions inside NVIDIA Isaac Sim as an alternative of testing solely inside a generic demonstration surroundings, VicOne mentioned.
The leaderboard from VicOneās Bodily AI Security Stress Take a look at CTF at DEF CON 34. Supply: VicOne
Radeis Extension for NVIDIA Isaac Sim now obtainable
VicOne asserted that Radeis provides robotics groups a sensible start line for exploring cyber-to-physical danger. It added that the free extension for NVIDIA Isaac Sim can assist engineering, security, and safety groups focus on danger via seen robotic outcomes slightly than treating vulnerabilities solely as summary software program findings.
For organizations requiring broader evaluation, VicOne Radeis is a pre-deployment cyber-safety validation resolution that assesses element, system, and AI mannequin dangers, then evaluates how vulnerabilities and assault eventualities may have an effect on the habits of the built-in robotic. The entire Radeis resolution helps deeper investigation, remediation planning, launch selections, and compliance readiness, mentioned VicOne.
The VicOne Radeis Extension for NVIDIA Isaac Sim is offered now as a free download for builders utilizing supported variations of NVIDIA Isaac Sim.
The submit VicOne releases free NVIDIA Isaac Sim cybersecurity extension based mostly on DEF CON 34 analysis appeared first on The Robotic Report.

