In November 2023 the GREYDIENT project organized a hackathon for its 15 PhD students. Divided in five groups of three they had two days to tackle one of two technical problems, develop a working prototype and present it as a business pitch highlighting core features and market targets.
The safe routing problem
The team is working for the honourable police officer Chief Clancy Wiggum. Together with Mr. Edward Finnerfield you are tasked to organize of the transport of the criminal mastermind Dr. Robert Underdunk “Bob” Terwilliger, Jr. The prisoner transport needs to be guided on its route through Springfield Munich. The route starts at the Marienplatz, where the court is located and goes to Schloss Nympenburg, where the accused will be placed in a maximum-security dungeon. There are two constraints to this route that need to be considered jointly: the probability that an accident occurs during the trip must be as low as possible, and the route time must be as short as possible.
Of interest to the stakeholders are any of the following (or combinations thereof):
• Identify optimal paths and alternatives, including risk estimates for each
• Confidence/uncertainty estimates on the proposed paths
• Interactive graphical representation of the paths and their uncertainties, possibly in the form of risk maps.
Relevant open source data and codes are available from the Unfallatlas and Openrouteservice.
Prototype demo
The student team composed of Damien Bonnet-Eymard (ESR02), Giada Collela (ESR09) and Juan-Pablo Futalef (ESR06) developed LIFESAVER to tackle this routing problem.
-
- Damien Bonnet-Eymard
-
- Giada Colella
-
- Juan Pablo Futalef
LIFESAVER is an innovative service for the transportation of high-value goods and people. We recognize that, even in the safest environments, risk exists and can severely impact the operation of the promised transportation service.
Our solution consists of a software suite that interacts with users to find the safest routes that vehicles should follow. We achieve this by exploiting historical data of accidents and analyzing multiple features that are more prone to cause accidents, such as maximum speeds, road types, and more. As a result, we obtain a rich granular characterization of the probability of accidents of roads, even in those where data is scarce, which enable us to minimize the overall risk within the route.
