16 screens, wheels, machine learning

2026

The flatbed trolley always seeks the shortest path between two points; the cost is sweat, memory, and the body’s wear. When the floor is smooth it seems to glide; workers learn instinctively to find such surfaces and avoid uneven ground. 

Sixteen trolleys occupy the exhibition space. Fifteen stand still. Their screens show reinforcement-learning-trained digital agents gliding frictionlessly through 3D reconstructions of real, vacant shops — moving at zero cost, optimising routes toward goods that no longer exist. Entirely free movement, entirely without meaning. The sixteenth trolley is motorised, and actually moves. It pushes forward, hits an obstacle, retreats, pushes forward again — a mechanical body trapped in an endless loop, searching for an exit that was never there. Real effort. Real friction. Real futility. 

Cargo that cannot arrive. Workers who cannot leave. Weightless digital flow and laborious physical effort alike collapse into different expressions of staying still — rehearsing immobility at two speeds within the same urban grid.