Playable City Sandbox: How (not) to get hit by a self-driving car

How (not) to get hit by a self-driving car By Tomo Kihara and Playfool (photo credit Luke O'Donovan)

A street-based game that challenges visitors to avoid being detected as a pedestrian in the eye of an AI. As surveillance cameras evolve and self-driving technology becomes increasingly prevalent, the cities we live in are starting to see us back. But how do these systems observe us, and where do their blind spots lie? This game welcomes anyone to challenge the AI and try to reach the goal without being detected. 

3 people in different poses. Layered over each person is a shape, label and percentage.

‘How (not) to get hit by a self-driving car’ is a street-based game that challenges anyone to avoid being detected by an AI-powered camera. When they join the game, players appear on a large LED screen, assigned with a detection percentage that rises as the AI recognises them more as pedestrians. If this percentage surpasses a specific threshold, the game ends for the players. In order to win, the players need to cleverly dodge the AI’s detection and aim to reach the goal undetected.

This game utilises a widely used object detection algorithm known as the Single Shot Detector (SSD), trained on a large-scale dataset. Every victory of a player exposes the inability of the system to detect pedestrians and highlights the flaws of these algorithms, which could potentially be applied to self-driving cars in the future. Upon victory, players are presented with a choice: they can opt-in to contribute the anonymised version of the image to improve the models or delete the image of themselves that were undetected by the AI. 

Through technology like surveillance cameras and self-driving cars, these advanced image recognition systems are quietly being installed in the public space without public awareness. By enabling anyone to play, the game aims to become an experiential platform to understand the workings of these systems, fostering dialogue about the challenges and potential dangers they present.

How (not) to get hit by a self-driving car showcase took place 3 – 7 July 2023 at Mshed.

The Team

Tomo Kihara is an artist and a game developer that designs and codes “toys for thought” — games and urban interventions that invite the public to explore critical questions through play. His recent projects have been nominated for the Ars Electronica STARTS PRIZE (2021) and exhibited at the V&A Museum (2022).

Playfool is a design duo comprising Daniel Coppen and Saki Maruyama. Together, they use play as a device to intervene in the relationship between humans and technology. Their past works have been awarded the Dezeen Award (2021) and exhibited at the V&A Museum (2022).

Recent Press

This street-based game challenges you to get hit by a self-driving car to outsmart its AI | Design Boom | 10.11.23

You win this street game by getting ‘hit’ by a self-driving car | FreeThink | 17.11.23

How (Not) to get hit by a self-driving car is a goofy yet sobering riff on modern technology | Kotaku | 12.12.23