SteerSense

Inspiration

With the rise of self-driving technology, a critical challenge emerges: how to program these vehicles to make ethical decisions in complex, real-world scenarios. To address this, researchers at MIT created MoralMachine.net, an interactive platform that presents users with thought-provoking dilemmas that autonomous vehicles might encounter. These scenarios, often involving difficult choices about safety and prioritization, highlight the ethical and moral considerations inherent in programming decision-making algorithms. By engaging with these challenges, we aim to push the boundaries of our understanding and test the limits of algorithmic decision-making.

What it does

SteerSense is an intelligent automated driving system that can tackle a wide range of driving conditions, ensuring safety and efficiency on the road. With SteerSense, drivers and pedestrians can rest assured that their safety is prioritized through advanced automation and reasoning abilities.

How we built it

We started by discussing various scenarios that a car can encounter while driving on the road. From there, we ultimately decided that the car will likely face two different types of situations: general driving and split-second decision making. Using these broad umbrellas, we reverse engineered the decision-making process for the car and came up with a set of rules that the car would have to abide to based on the scenario. After thoroughly defining the constraints in the scenario, s(CASP) was able to recognize the correct course of action.

Challenges we ran into

• Taking a logical approach to ethics
• Establishing proper constraints to ensure one course of action
• Incorporating a web of parameters that accurately summarize complex situations
• Defining project scope

Accomplishments that we're proud of

• Two model integration
• Accurate ethical simulation
• Support for lots of different parameters

What we learned

• A new language, prolog!
• Addressing a complex issue with seriousness and depth
• Translating human reasoning into code

What's next for SteerSense

• Collaboration with Regulatory Bodies: Self-driving cars are relatively new technology and they will likely face lots of regulation in the future. We want to be collaborating with regulatory bodies to ensure our code is up-to-date with the latest regulation
• Human vs Machine: Our model currently assumes the Machine will be driving in all cases; however, it is sometimes better for a human to be driving.
• In-Car Ethical Assistance: Self-driving cars currently have a black-box type model that can raise concerns around transparency. By providing ethical assistance, the driver can be actively informed as to why the car is making certain decisions.

Built With

• ciao
• linux
• prolog
• s(casp)
• wsl