The proceedings from the 11th International Conference on Ubiquitous Computing held in late September in Orlando, Fla., included a paper from five University of Washington (UW) researchers titled, “A Spotlight on Security and Privacy Risks with Future Household Robots: Attacks and Lessons.” The paper describes a study that has major implications for the consumer robotics market. The research indentified a shocking lack of even rudimentary security and privacy safeguards within a number of popular household robots. To address the problem, the UW group formulated a set of core questions that could be used to ensure that privacy and security issues are addressed during the development of consumer robotics products.
The University of Washington paper described a multipart study that began with an examination of three common consumer robots for security and privacy vulnerabilities. Based on experimental results, the researchers then identified the key security failings common among the products. The researchers concluded the paper with a set of questions that illuminate how the operational properties of consumer robots under development could impact the security and privacy of the product’s users. The robots evaluated were:
- Spykee. Spykee is a tracked, mobile robot released by Erector/Meccano in 2008. The robot, which is targeted to children and sold as a smart toy, is controlled wirelessly via Wi-Fi (802.11) by a computer or over the Internet. The operator moves Spykee using a webcam for guidance, and can take pictures and video, as well as listen, speak through the microphone, or play various built-in sounds
- Robosapien V2. Launched in 2005, Robosapien V2 is the second generation of WowWee’s Robosapien line of robots. The legged, mobile robot, which is targeted to children, is controlled via an infrared controller. The sensor-laden robot has a fully articulated body that can be controlled programmatically (very limited), as well as by a number of preprogrammed functions that can be executed sequentially. It also contains a color camera, infrared vision, and audio sensors.
- Rovio. WowWee’s Rovio, released in 2008, is a Wi-Fi based mobile webcam targeted to adults that can stream back live video and audio. Rovio can be controlled wirelessly via 802.11 either locally or remotely using any Web-enabled device including a PC, cell phone, smartphone, and PDA.
Security Negligible or Nonexistent
The UW researchers discovered that basic security measures for household robots were negligible or nonexistent. Reliance on home wireless networks for control and monitoring opened another attack vector. Remote control and viewing features across the Internet were implemented with little adherence to security best practices. In short, household robots designed to capture and relay audio and video from their environment, and to react to remote commands, included no security awareness or safeguards in their design.
The researchers also found that the security deficiencies extend beyond the products and network services. For example, robot vendors that provide proxy services for connecting to their products do not address their responsibilities concerning the information they forward. Are the video and audio streams recorded or cached at their site? If so, what are their security procedures for access to those streams? Are such recordings available to law enforcement?
While the paper lists various harmful scenarios (minor vandalism, tripping those with limited mobility, etc.), the most common consequence of the lack of security is spying. Intercepting the audio and video feed from consumer robots, especially those such as Rovio, which are specifically designed for surveillance, provides outsiders an inside view of everything that happens near the robot. Additionally, since security/surveillance systems like Rovio stream audio and video to other network-enabled devices such as video game consoles, cell phones, and smartphones, a lost or stolen phone would give eavesdroppers access to the owner’s home.
To date, there has been no report of malicious use of a household robot. However, University of Washington research makes it clear that makers of consumer robots must provide basic security and privacy safeguards in their products going forward. The researchers correctly note that as consumer robotics products proliferate and increase in functionality, especially the class of security and surveillance devices that can stream audio and video feeds from the home, security and privacy safeguards must be increased and become more robust.
For consumer robotics to enter the mainstream, producers must include security and privacy controls in their products that are on par with computer software used in the home and in consumer electronics devices such as cell phones. Indeed, given the unique ability of robots to move through their environment and physically interact with the people and objects in it, security and privacy safeguards must exceed that of other consumer electronics devices. To assist robotics developers with this difficult design task, the UW group created a set of questions that generates a risk model that can then be used to develop specific privacy and security controls.
The Bottom Line
University of Washington researchers conducted a formal study of the security and privacy safeguards in three consumer robotics products and incorporated their results in a paper for a conference on ubiquitous computing. They concluded:
- Basic security measures commonly found in personal computers and other consumer electronics devices were lacking in the household robots.
- Security and privacy issues must be addressed before the household robotics market can grow as expected.
- Robots are mobile and physically interact with people and objects in their environment. Therefore, security and privacy safeguards incorporated into robots must exceed that of computers and other consumer electronics devices.
- It is possible to create a privacy/security risk model by determining how a robot’s actions might affect the security and privacy of users and their property, and then use the risk model to develop specific privacy/security controls in household robots.