When an Uber car picking up passengers is a robot, passengers want assurance that the ride is going to be affordable, efficient, smooth — and safe.
That’s where researchers like Neeraj Gandhi, a doctoral candidate in Computer and Information Science (CIS) and scholar at the Penn Research in Embedded Computing and Integrated Systems Engineering (PRECISE) Center, come in.
Gandhi focuses on improving the safety and security of networks of computers that collaborate to control physical devices, such as self-driving cars. He also looks at these processes in multi-rotor aerial drones, which can be used for agriculture, mining, mapping, surveillance and intelligence, and system security for tasks performed by multiple robots.
Gandhi first became interested in Cyber-Physical Systems (CPS) in high school, where he worked on building a simple robotic arm controlled by muscle electrical signals. Since then, he has worked in a number of different areas and found himself drawn to the intersection of different fields.
Gandhi, who received his master’s in robotics at Penn Engineering and his bachelor’s in computer engineering at the University of Virginia, has also researched extraterrestrial mining robots, body sensor networks for dementia agitation detection and prediction, and photoacoustic imaging for improved surgical guidance in traditional and robotic surgery. His advisor is Linh Thi Xuan Phan, Associate Professor in CIS and a member of the PRECISE Center.
With research spanning a range of systems and locales, including vehicles, factories and robots, Gandhi’s goal is for real-world practitioners to use his research to make their AI processes safer and more tolerant of the faults that will inevitably occur. “It is practically impossible to prevent systems from undergoing any fault, so the most we can do is ensure that when faults do occur, they are guaranteed to be handled safety,” Gandhi says.
Gandhi’s most recent research project helps AI systems recover from a fault using fewer computing resources, which keeps the system safe but also allows it to run longer and more efficiently.
“This research enables lightweight security in systems like cars, which currently operate over unsecured communication mediums,” Gandhi says. “Designers concerned about the resource demand of techniques that were developed for distributed computing systems employed in data centers now have the option to use this less resource-intensive approach, while still being able to provide guaranteed protection against a broad set of benign and adversarial faults.”
Connecting systems in networks has enabled innovations like self-driving cars, but these systems’ complexity means they can easily experience bugs or be targets for attacks. Gandhi’s research focuses on how the systems can better respond, paving the way for an expansion in how the systems are deployed.
“Neeraj has developed entirely new approaches and out-of-the-box solutions to address important challenges in security, safety and real-time for modern cyber-physical systems and distributed robotic systems. Among many other contributions, his research provides a way to build cost-effective and efficient large-scale complex CPS that can self-adapt in real-time and that are provably resilient to faults and security attacks,” says Insup Lee, Director of the PRECISE Center.
“The solutions Neeraj proposed often integrate and extend techniques from multiple disciplines in a novel way, from cyber security and distributed systems to real-time scheduling and control systems,” Phan adds. “The results also open up many new and exciting research directions at the intersections of these domains that will play a critical role in enhancing safety, security, adaptability and performance for modern real-world systems.”
Another of Gandhi’s projects focuses on multi-robot systems where several drones come together to form a larger one. One challenge of such systems is that when one or more rotors on one of the drones fails, it can compromise the safe operation of the entire system. Gandhi’s research identifies methods for more easily finding faulty rotors and having the system take action to compensate for the fault so movement isn’t impacted. One such method is a self-reconfiguration technique that has the larger structure break down into smaller pieces and re-assemble into the same shape, but with the faulty rotor moved into positions of low impact on the motion of the system as a whole.
“Our techniques do not require any hardware changes and are computationally lightweight, and thus well suited to run on simple embedded systems,” Gandhi says.
Gandhi’s final and ongoing project focuses on networked multi-robot systems, which are increasingly common in today’s world, including in warehouses. When these systems experience failures, it can cause major problems – for example, one robotic grocery packing company has had several robot safety issues that caused millions of dollars in damages over the past few years. Furthermore, as the scale and frequency of use of these systems grow, cyberattacks become more likely.
Gandhi’s research on this topic focuses on having a compromised or “buggy” robot automatically go into a “safe mode” for a bounded amount of time. This involves having the system self-configure to cut out a faulty robot, which the system identifies using the “tokens” that each robot in the network collects from nearby machines certifying that they are operating correctly.
“There has not typically been much crossover between the distributed systems research and robotics research,” Gandhi says. “Insights from building safety and security into data centers or related systems can be used to build lightweight safety and security approaches for distributed robotics.”
During his time at Penn, Gandhi greatly expanded his knowledge of real-time systems, scheduling, security, and fault detection and recovery in distributed systems. After graduating this summer, Gandhi will be joining Oracle’s Exascale system team, working on cloud systems optimized for running the Oracle Database. His future work is specifically in a new type of distributed file system to be deployed in cloud systems that have tight hardware-software integration, where the knowledge gained in his time at PRECISE will be particularly useful.