Jace Sheu, Ishan Tikku, Ruila Puskas-Juhasz, Liwen Zhang, Valerie Huang, Asif Mohammad
Using historical crime and safety statistics, NavSafe uses machine learning to ensure safe and efficient navigation experience in our cities.
Berkeley, California – Traveling in a big city such as San Francisco without being well informed about safety of neighborhoods at different times of the day might be extremely risky for commuters. With no apps addressing this challenge of tracking real-time safety of a route, this group of UC Berkeley graduate students decided to build an app that can be your trusted safety advisor when you are traveling in new cities. With the core mission to develop a safe navigation service by adding context about crime statistics and safety reviews of neighbourhoods, this team uses machine learning to predict crime in different neighborhoods and recommend safe routes in San Francisco city.
The team developed an algorithm that carves the city’s neighbourhoods into sectors and labels them as safe and unsafe ones during different times of the day and night. Utilizing various clustering algorithms, this group of UC Berkeley students developed a navigation route recommendation model that takes into account severity of crime, time of the day and perceived neighbourhood safety. For a person travelling from point A to point B in the city, thanks to NavSafe they can now be guaranteed of a safe and secure commute. NavSafe’s recommended routes avoid unsafe sectors and empower travelers with safety information that a local would know.
NavSafe received positive feedback from users who found immense value from these alternative safe route recommendations during their recent travel in SF. The team intends to build on the feedback received for SF city and scale their service to 10 metro cities in the US in Phase 1. The team also observes that their roadmap will address challenges such as biases in crime data sources and potential downsides of the alternative route recommendations that might adversely affect small businesses that might lose pedestrian traffic.