TrakBot -- Allison, Han, Jeremy, Kate

This project is interesting to work on but has relatively few real-world uses (trail mapping, autonomous snowblowing) outside of ethically flawed applications (autonomous military robotics, surveillance, etc). We’re excited to work on this project for personal interest in the components and development, but would need to give it far more consideration before working on a similar project in a corporate or funded-research setting.

1. Directly facilitate injury to living beings. For example: could it be integrated into weapons or weapons systems?
• Yes. It is possible to attach weapons systems and other devices capable of harm to this robot if a party in ownership of the robot decides to. The kit robot we built off of is available for purchase by the public, the microcontrollers and required iOS device are off-the-shelf, and all code is open source. Thus, it is possible for this project to be reproduced and expanded for harmful purposes.
2. Raise safety, privacy, or security concerns.
• Potentially. In our system, data including camera images from the iOS device is collected but only streamed live through an access point on the Raspberry Pi to be used by our autonomous navigation system. However, it is possible for the data to be streamed over a wireless network and accessed by other computers with the correct credentials. As is the case with working with any camera images, this creates the potential for breach of privacy of a person’s identity and/or location.
3. Raise human rights concerns.
• Unlikely. As is, the system is fairly unbiased since it is working with trails in woods and geographic location, with no humans involved. However, as mentioned in point (1), and as is the case for any open source autonomous platform, our system can be misused and repurposed for discriminatory and harmful uses.
4. Have a detrimental effect on people’s livelihood or economic security.
• Potentially. Because this robot is capable of autonomous navigation outdoors, a space that is less automated than indoor spaces such as factories and warehouses, there is potential for it to displace people who work outdoors in jobs such as trail maintenance.
5. Develop or extend harmful forms of surveillance.
• Yes. This is an autonomously driving robot with an onboard camera that also has GPS and local positioning via visual inertial odometry functionality. As a result, it is possible to exploit any and all of this data.
6. Severely damage the environment.
• Potentially. This robot can physically damage the environment that it drives through, and depending on what is attached to it, it could do increasing harm. In driving the robot around, we did accidentally damage many plants both in autonomous mode and manual control.
7. Deceive people in ways that cause harm.
• Unlikely. The robot does not have any communication systems that could be used for fraud, impersonation, etc. It is not attempting to imitate humans in any way.

As we are sourcing data for our machine learning model, we need to be conscious of what data we are using and how we are getting it. As we are not working with data involving people, we have less concerns about bias as our model will have no human impact. Our data was all collected by researchers working in the same field as us, and was published for open source use so we are free to continue using it.