TravelSafe

sample screenshot of the conversation with assistant

Inspiration

In many metropolitan cities, one would find public transit systems, whether they are buses, subway systems or streetcars. When we see people engaged in altercations, such as physical or verbal altercations that could put the public at a risk, a good person would typically call for help. In most cities and regions, it is not possible to text 911/other emergency services numbers to receive help, making phone call the only manner to ask for help. However, because calling for emergency services (ex. the police) requires one to physically pick up the phone and report the crime, this may put the caller in a position of possible danger. In some cases, a bystander may feel uncomfortable contacting the police because of fear of intimidation and injury from the perpetrator.

We created a chatbot that would effectively allow users to report crime in a discreet manner, such that a crime could be reported by typing rather than speech. This would promote more people to report crimes and create a safer environment for all citizens.

What it does

TravelSafe is a chatbot which utilizes Actions on Google and Google Cloud Services to complete what a person would normally do in an emergency situation requiring emergency attention, which is call for assistance. The user can type in the information about the situation and locations and send it to emergency services without needing to speak a single word.

How we built it

We built TravelSafe using Actions on Google, Firebase, Node.js, Google Cloud Platform, and Twilio IP.

Challenges we ran into

One of the most significant challenges we ran into was the fact that error messages and logs were not in the most obvious of places. This led to many headaches and trial-and-error attempts to fix problems that arose. Eventually, through experience, we were able to find out where all logs were and they were able to fix them right away. We also ran into challenges implementing MongoDB (we later found out that the documentation was incorrect), which limited our project scope.

What's next

We want to be able to eliminate bias inside the application and prevent unneeded calls from being dialled. In other words, some people may report situations where they are not necessarily in danger but due to personal bias. For example, some person might report a person walking down the street wearing a big hood because they feel unsafe, but that person may not be a dangerous person. We also want to prevent children from dialling into the station, because there are many instances where children have dialled emergency services complaining about non-urgent manners, crowding the phone lines.

Please see the alpha deployment in link.