Inspiration

If you were walking down Wyoming St in Detroit, Michigan, or 1st St in Jackson, Tennessee, you have reason to be terrified. You have a 10% chance of falling prey to violent crimes such as armed robbery and rape. In the “Land of the Free”, people ought to be free to tread quiet streets at night without fearing for their lives.

How can we improve the safety of people on America’s most dangerous streets? How can we shorten response times when a violent crime occurs? How can we make people feel safe and supported while walking alone at night?

At StreetSafe, we are seeking answers to these questions. Thus, we have created a virtual buddy that accompanies people on late-night walks and mobilizes for help when signs of violent crime are detected.

What StreetSafe Does

StreetSafe is a virtual buddy app that accompanies you on late-night walks and alerts effective responders (i.e. police, people nearby) if any anomalies are detected. StreetSafe has the following unique features:

• Reliable Buddy: Powered by IBM Watson, StreetSafe tells stories, cracks jokes and sustains conversations for the entire duration of a walk. At the same time, it serves as a trusted guide—advising users on the safest path to take according to real-time crime information.

• Anomaly Detection: The app listens for sounds of distress (e.g. screams, danger words and long pauses in response). It also detects sudden acceleration and impact of a phone, which might indicate a snatch theft.

• Emergency Alerts: When something suspicious occurs, StreetSafe will ask a user thrice if everything is okay. If no response is given, the app will send out text alerts to effective responders. These include law enforcement agencies and anyone within a 1-mile radius of the incident site (who will receive an SMS). Information on the location and likely type of incident (e.g. snatch theft, kidnapping, etc) will be provided, thus facilitating quick mobilization of resources to help the victim.

How We Built It

StreetSafe is conceptualized on Adobe XD Creative Cloud, and built on Android Studio with Java. We used sound level monitoring in Java to pick up any unusual voice modulations (e.g. loud screams).

To power our conversational feature, we personalised IBM Watson to respond to specific comments or requests (e.g. Tell me a short story). We used Twilio REST API to send SMS alerts to people within a 1-mile radius of a crime scene.

Challenges We Ran Into

We tried to use Firebase to provide a Google login into the application, but we got stuck when the SSA1 key verification was failing continuously. It was also complex to figure out how to use sound libraries in Java.

Accomplishments that We Are Proud Of

Our virtual buddy is able to hold sustained and engaging conversations with users. It goes beyond exchanging pleasantries to telling stories and cracking jokes.

Also, we are proud to have created the idea of mobilizing potential good Samaritans near a crime scene to assist a victim. Other apps on the market focus on alerting law enforcement, even though bystanders can be a great crime-fighting force.

What We Learned

In the technical aspect, we improved our proficiencies in Java and IBM Watson, while learning how to solve problems when a particular tool did not work. In the team aspect, we learned how to combine our diverse skills in a complementary way, with the common goal of creating tech for good.

What's next for StreetSafe

We plan to implement the following in the next six months:

• Build an ML model to classify different sound profiles better, in order to ascertain distress signs with greater accuracy.

• Use gyroscope to detect sudden acceleration the phone, such as when the phone is snatched or dropped on the ground.

• Create three different threat levels—high, medium and low—based on historical data on crime in an area and vary sensitivity to distress signs based on threat levels.

• Enable virtual buddy to process real-time crime data and advise users to change routes.

• Perform a trial of the SMS alert system on fifty participants in a controlled environment.

Projected Use Case

Betty wants to go from her workplace to her friend’s house at a Tuesday night., but is concerned as the streets are dark and quiet. She pulls out her phone and opens StreetSafe. She types in her friend’s address and starts a conversation with her familiar virtual buddy, Lisa. Betty tells Lisa about her day at work, before asking Lisa to tell her a motivational story.

The walk is projected to last 20 minutes. All of a sudden, a man comes up to Betty and grabs her phone. Betty is too shocked, and stands transfixed and speechless as the man tearing down the street.

Detecting the sudden acceleration in Becky’s phone, Lisa asks Becky, “Becky, are you alright?”. After asking thrice with no response, StreetSafe sends a text alert to the Police, informing them of a suspected snatch theft. An SMS alert is also sent out to everyone in a 1-mile radius of the crime scene:

“ Dear Sir/Madam, A suspected snatch theft has occurred at 2431 21st Street, which is in your vicinity. We urge you to exercise extreme caution in the area, while rendering assistance to the victim if you are able to do so. Thank you.”

George, who is walking along the street a block away, checks the SMS, looks up and sees a man sprinting towards him. “Stop! Stop!” he yells, but the man refuses to stop. George, a running enthusiast, gives chase to the thief and manages to pin him down with the help of other passers-by 5 minutes later. Soon enough, the police arrive at the scene to apprehend the thief, and Becky gets her stolen phone back.

A violent street crime is thwarted, thanks to Lisa’s alertness and ability to mobilize the community to come to Becky’s aid.

Built With

Share this project:
×

Updates