FAQ
Photon Echo delivers labeled synthetic datasets generated from simulated scenarios. Depending on the project, outputs can include RGB images or video, depth, segmentation, masks, bounding boxes, pose labels, body state labels, metadata, manifests, and label schemas.
Photon Echo uses simulation to generate data, but the core deliverable is synthetic data. Source scenes or simulation assets may be included when needed, but the primary product is dataset generation from simulated scenarios.
Photon Echo is not positioned as a frontier foundation model lab. The company is moving toward the environment, scenario, and evaluation layer that physical AI systems need: model-ready environments, scenario-driven synthetic data, and repeatable outputs for training and testing.
The dataset starts from a real world condition the customer needs to cover, such as poor lighting, occlusion, unsafe proximity, unusual body position, weather, clutter, or a rare object state. The scene is then varied in controlled ways to generate usable training or test data.
No. It extends coverage and supports testing where field collection alone is slow, unsafe, or incomplete.
Yes. Photon Echo datasets can support validation, benchmarking, perception testing, and safety review, but the core deliverable is the dataset itself unless a separate reporting scope is agreed.
Photon Echo serves robotics, autonomy, inspection, warehouse automation, healthcare, public safety, and physical AI teams that need more coverage for rare conditions.
