MirrorSpace is an early-stage physical-AI company developing a multimodal spatiotemporal perception and memory stack for robots operating in changing environments.
About
MirrorSpace is developing an embodied spatiotemporal intelligence stack that combines multimodal perception, scene memory and prediction. Its MirrorSense V1 prototype combines RGB, LiDAR and thermal inputs to create dynamic representations of physical spaces for downstream robotic control.
Founded in 2026, the Dongguan company is moving from prototype work toward early deliveries. Its September Seed+ round is funding perception-hardware commercialization, continued spatial-memory development and training of a world model for changing real-world environments.