Crowdtesting for Milestone: a realistic AI

"The collaboration with UNGUESS has allowed us to have a large user base in a very concentrated time, thus obtaining very useful feedback to be integrated into the title in development."

IVAN ORVIETO Head of Development Operations

Context

With the video game RIDE 4, Milestone's Artificial Intelligence (AI) agent makes its debut and thanks to advanced technology manages to replicate the finest details, shattering the barriers between game and reality. The rival riders within the video game succeed in managing complex aspects: they are influenced by the weather conditions, the duration of the race, they decide when to stop at the pit stop, how much fuel they need and remember previous races in order to develop new strategies. To make this game an unforgettable experience and ensure it lives up to expectations, the behaviour of the riders must appear extremely realistic, and the dynamics of the game modes must be respected.

Objectives

• COMPATIBILITY: perfect operation on a high number of hardware configurations of both dynamics and graphics.
• AI REALISM: the rider s behaviour must be realistic in every situation.
• RESPECT OF THE GAME LOGICS: for example, the actions of the mechanics who must correctly respond to the rider s requests, the championship that is unlocked only when a certain number of experience points have been obtained or previous races have been won, etc

The solution

Crowdtesting campaigns focused on verifying:

1. specific complex test cases connected to the use of the RIDE 4 video game;

2. the proper functioning of the AI (Artificial Intelligence) engine in dozens of hardware configurations in which the game will be installed (mobile, PC and console).

Results

• 100+ Hardware Configurations
• 100+ Testers fresh eyes working simultaneously
• 3x Clusters with different objectives (AI, game logic, career objectives)
• 389 Bugs reported
• 24 Critical bugs identified
• 87,5 working days of testing reduced to 15 days