State-Manifold Science: one object, a brain and a cerebellum, connected.
Recorded demonstration, October 9, 2026
Screen recording of the LD-50 laboratory prototype's live console (53 s, no sound). Staring at one cell: range and message are decoded from the same frames; the received text matches the transmitted text bit for bit (35/35 slots) with 0 wrong frames delivered. Continuous scan: the measured 3-D point cloud and the tensor potential field drawn as a 3-D energy landscape (peaks = obstacles, valleys = passable) while the link keeps running. Conditions: indoor, one direction, the device receives its own returned beam, range validated 1–6 m, about 3 kbit/s delivered; research prototype. Download (MP4, 5 MB)
V131 Lab · V131 Institute for State-Manifold Science
State-Manifold Science studies one object: the state Ψ, the complex wave function of a light field, together with its history. We work on it from two sides. The brain is a unified-field model, built from the theory of that object. The cerebellum is physical AI: our second-generation machine, in which one laser beam senses, communicates, and controls at the same time. Our goal is to connect the two.
Gather the scattered volumes into one; keep the two ledgers apart. Forward computation is the method; mismatch is the teacher.
1 · The machine: native ISAC, measured on the bench on September 20, 2026
150,000
frames processed in about ten minutes, none lost
99.907%
of frames accepted for delivery (149,860)
0
errors observed in the delivered, decoded data
2.37 m
median range from the same stream, with a spread of 2.16 cm (median absolute deviation)
First-generation LD-50 laboratory prototype, indoors, one fixed direction, repeating payload, with error correction and a receiver confidence check. The uncorrected signal still contained errors (raw bit error rate 0.064%). Full conditions and limits → · Technical report on Zenodo (DOI 10.5281/zenodo.22904657)
Native, not glued: the manifold is the payload
The beam that measures is the beam that communicates.
A glued system pairs a laser rangefinder with a separate laser link and has to keep two narrow beams, two timings, and two receivers on the same spot. In native ISAC there is only one beam, one received signal, and one timing reference. The data is the shape of the transmitted signal itself, and the measurement is read from that same shape, so the two do not disturb each other.
The next stage is native ISCC: Integrated Sensing, Communication, and Control, in which the same stream also feeds the control of drones, robots, and vehicles. Control has not yet been validated in hardware.
2 · The second-generation machine: where it stands
- What it is. The second-generation machine is not a radar or a lidar. It is a physical-AI cerebellum: one beam senses, communicates, and computes the control quantities itself, and it plugs into any carrier, whether a drone, a robot, a vehicle, or a fire monitor.
- Hardware. It is a coherent optical design built around a telecom coherent module, with a four-channel, dual-polarisation receiver on one board (Y18). As of November 2026 the board is in manufacturing.
- First measurements. Once it is on the bench, we will report the first three optical-link numbers here. Their pass and fail criteria were registered before power-on, and the results will be published whether they pass or fail.
- Until then, every second-generation figure on this site is a design target (projection), not a result.
3 · Application: fire and rescue (projection, design targets)
Sensing, communication, and control for firefighting drones
Holding position next to a burning tower, with no satellite fix.
Between glass towers, satellite positioning is unreliable and cameras see reflections, smoke, or nothing at all. A heavy-lift firefighting drone also has to hold steady against the recoil of its own water jet. The advantages of a coherent 1550 nm beam follow from first principles:
- Velocity read directly. Each return carries its own Doppler shift, so speed is measured, not differenced from positions.
- Position measured against the facade itself, not against the sky.
- Through smoke. 1550 nm light passes through smoke, and coherent detection ignores flames and background light.
- Glass cannot fool it. Returns from a wall and from a reflection in glass differ in phase, polarisation, and motion.
- Many drones, one field. More drones means more measured points in a shared field, not more interference.
These are design targets. They will be replaced by recorded results once the second-generation machine has flown. Full application page (Chinese) →
Work with us
See it live
We demonstrate the first-generation laboratory prototype in operation, at our laboratory or live over video: data and distance recovered from the same optical stream as it happens.
Fire and rescue
Fire services and drone makers working on high-rise firefighting and rescue are welcome to discuss trials with the second-generation machine.
Research and industry
Groups working on optical communication, lidar, integrated sensing and communication, drones, robotics, or vehicles can see the prototype running and discuss a licence.
What is still being tested
- A separately transmitting and receiving pair of devices. The first-generation bench results used the prototype's own returned signal.
- Fresh, non-repeating data. The continuous tests replayed a fixed table of codewords.
- Outdoor distances, moving targets, sunlight, and temperature changes. Current distance evidence covers up to about 6 m.
- Absolute range calibration. An unresolved scale offset of about 2.8% remains at the first-generation operating point.
- The Control stage of ISCC, and every second-generation figure.
The institute's work
Native ISAC results
Bench measurements, communication reliability, and range from one shared signal.
The physical-AI world model
GPU measurements, model-based control, and checks on how reliable the model's predictions are.
Knowing when to abstain
Preregistered criteria, negative results kept on record, and a protocol that withholds a decision when its criteria are not met.
Publications and licences
Defensive publications on Zenodo, record fingerprints, and licensing terms.
Devices and photonic chips
The photonic chip
The silicon photonics architecture, verification work, and clearly labelled projections.
Photonic cipher validation
Recorded fiber-link tests, latency measurements, and responses to spoofing attempts.
Photonic computation
The proposed computation and readout approach, its hardware basis, and its limits.
Theory: State-Manifold Science
The longer-range ideas behind V131. These pages are exploratory; each states its level of evidence and its limits.
Information dynamics
The concepts behind V131 and the proposed links between information and physical systems.
Cosmology
A speculative framework for black-hole cores, dark energy, dark matter, and neutrinos, with testable predictions and stated limits.
Design comparisons
Physical-invariant methods compared with statistical approaches.
How to read the results
Measured means a recorded test under stated conditions. Simulation means a software experiment. Projection means an expected performance or proposed application. These are different levels of evidence; read the conditions alongside each number.