Perception
IN PROGRESS
PHYSICAL SYSTEM
- Sensor interfaces
- Camera and inertial test inputs
- Timing and calibration
SOFTWARE SYSTEM
- Vision pipeline
- Detection and tracking
- Uncertainty handling
Autonomous systems that can observe, understand, and keep working in remote environments.
Antshiv Robotics is developing the software and compute foundations for autonomous sensing across biodiversity conservation, infrastructure inspection, mapping, search and rescue, public safety, and defence research.
Today we are building flight-control mathematics, simulation, onboard CPU AI, and portable compute infrastructure. Over the next two to three years, we plan to connect these components into integrated aircraft and ground systems as financing, hardware access, and field partnerships develop.
RESILIENT AUTONOMOUS SENSING · OWNED COMPUTE
Observe remote environments
Support conservation, inspection, mapping, public safety, and reconnaissance with adaptable sensing systems.
Run useful AI locally
Use compact CKE models when connectivity, power, latency, or data ownership makes remote inference impractical.
Add portable CPU compute
Move larger models and mission analysis onto expandable ground nodes without making basic operation depend on them.
Carry the evidence forward
Preserve tests, telemetry, model provenance, measurements, and field observations for the next engineering cycle.
We build compiler-driven systems from the web interface to AI runtimes and flight control.
Antsand turns structured content, data, design, and site operations into deployed web applications. CKE turns model circuits and numerical contracts into auditable C runtimes for inference and training. Both make the path from intent to execution visible instead of hiding it behind a large framework.
The same method continues into flight control: begin with quaternion mathematics, state estimation, and control algorithms, then implement and measure them in C for the processors, sensors, power limits, and aircraft Antshiv Robotics selects. Across the stack, we optimize for auditability, performance, and the engineering intuition to understand why the complete system behaves as it does.
HOW THE WORK CONNECTS
Start with mathematics and constraints
Describe motion, sensing, uncertainty, computation, timing, energy, memory, and communication before choosing an implementation.
01 / MODEL
Decide where each responsibility belongs
Separate flight safety, onboard perception, ground compute, data, and operator interaction into interfaces with known failure behaviour.
02 / ARCHITECT
Turn the architecture into working systems
Implement control software, C kernels, model circuits, simulation, telemetry, data pipelines, and the hardware paths connecting them.
03 / BUILD
Use evidence to guide the next design
Run the system, compare it with mathematical and software references, inspect failures, and carry the measurements into the next engineering cycle.
04 / MEASURE
Flight control + CKE + CPU systems + Antsand
Flight-control algorithms turn estimated state into safe action. CKE turns model circuits into auditable CPU execution for inference and training. Expandable CPU nodes provide additional memory and compute, while Antsand organizes the interfaces, evidence, telemetry, content, and operational records around the work.
THE CONNECTED STACK
How model inputs become measured CPU execution
CKE provides the AI execution layer inside the larger Antshiv system. It connects model structure to explicit circuits, generated C, native CPU kernels, and evidence we can inspect.
CPU AI · ONE LAYER OF THE SYSTEM
Three programmes, one engineering method
Equation → model → numerical contract → kernel or controller → native execution → hardware behaviour → measurement.
C-Kernel-Engine
A C-first compiler and runtime that turns model weights and circuit templates into inspectable generated C and native CPU kernels.
Flight control and estimation
Rigid-body mathematics, state estimation, simulation, and sensor fusion built methodically toward controlled autonomous systems.
Antsand and evidence
Structured experiments, telemetry, evidence dashboards, deployment tooling, and technical publishing that preserve the reasoning behind each result.
Building toward a resilient aircraft and portable CPU ground system.
Antshiv Robotics is bringing its flight-control, CPU AI, and systems work together as one long-term aircraft programme. The flight computer handles stability and recovery, CKE adds onboard perception, and portable CPU nodes provide larger models and mission analysis when a network connection is available.
This is a two-to-three-year development direction. Current investment is concentrated on the software layer and expandable CPU nodes. Aircraft integration follows as financing, hardware access, and test partnerships allow us to move from simulation into hardware-in-the-loop and relevant-environment testing.
RESILIENT UAS · SYSTEM ROADMAP
Flight control and recovery
Stabilization, state estimation, navigation, geofencing, return-to-home, and emergency recovery belong to the deterministic flight computer. This gives the aircraft a reliable operating foundation before additional AI capabilities are introduced.
01 / SAFETY FLIGHT COMPUTER
Compact models add local perception
A power-bounded companion computer runs selected vision, audio, or language circuits for perception and decision support. A model crash or missed deadline cannot take ownership of the safety loop.
02 / ONBOARD CKE COMPANION
Larger models are available when the link is
Expandable CPU nodes provide more memory, storage, and distributed compute for mission analysis, multi-aircraft coordination, and larger models. Loss of that link reduces capability, not basic flight safety.
03 / PORTABLE CPU GROUND NODES
How we plan to integrate the system
We will move from component tests to closed-loop simulation, hardware-in-the-loop testing, and relevant-environment trials. At each stage we will measure latency, power, communication behaviour, recovery paths, and the provenance of models and observations.
INTEGRATION ROADMAP · CURRENT WORK TO FIELD TESTING
- NOW · Component mathematics, CKE runtime paths, simulation, and test infrastructure
- NEXT · Define the interfaces between flight control, onboard inference, and ground compute
- THEN · Closed-loop simulation and hardware-in-the-loop testing
- RELEVANT ENVIRONMENT · Test an integrated prototype under cold, wind, vibration, and constrained power
- FIELD WORK · Supervised trials for inspection, mapping, search and rescue, and situational awareness
What we have built and measured so far
These examples link to the source code, test reports, and engineering records behind the current work.
Multi-family model bring-up
CKE supports multiple Qwen, Gemma, GLM, Nanbeige, and multimodal paths through explicit model-family contracts.
Generated C that fails closed
The v8 pipeline lowers model circuits to deterministic C and rejects missing kernels, routes, or numerical contracts instead of silently falling back.
Cross-runtime numerical parity
Layered scalar, ISA, and end-to-end gates compare CKE with PyTorch and llama.cpp and identify the first divergent operation.
CPU execution across real hardware
Current work spans AVX2, AVX-VNNI, AVX-512, AMX BF16, and ARM NEON, with measured support stated separately from work still being hardened.
Simulation and control foundations
Rigid-body mathematics, control systems, sensor models, and reproducible simulation are being developed before any autonomous-product claim.
A public engineering record
ShivasNotes publishes the derivations, architecture, debugging evidence, and limitations behind the work.
Autonomy is a chain of hardware and software contracts
These are research responsibilities and maturity targets, not a claim that a finished autonomous drone is currently available.
State and navigation
IN PROGRESS
PHYSICAL SYSTEM
- Rigid-body dynamics
- Actuator and vehicle models
- Physical constraints
SOFTWARE SYSTEM
- State estimation
- Sensor fusion
- Path and mission planning
Energy and endurance
IN PROGRESS
PHYSICAL SYSTEM
- Rotor and power models
- Battery and payload limits
- Measured test inputs
SOFTWARE SYSTEM
- Energy budget
- Mission constraints
- Range estimation
Safety and verification
IN PROGRESS
PHYSICAL SYSTEM
- Test constraints
- Benchtop and simulation rigs
- Failure instrumentation
SOFTWARE SYSTEM
- Deterministic simulation
- Parity and regression checks
- Fail-safe logic
Field sensing
PLANNED
PHYSICAL SYSTEM
- Payload interfaces
- Environmental sensors
- Edge compute
SOFTWARE SYSTEM
- Telemetry records
- Evidence pipelines
- Human review
Remote deployment
PLANNED
PHYSICAL SYSTEM
- Docking and charging
- Networking
- Serviceable hardware
SOFTWARE SYSTEM
- Long-horizon autonomy
- Fleet coordination
- Operational monitoring
The engineering record is part of the product
Each article turns a code change, mathematical question, or measured failure into a durable explanation.
Templates are circuit maps
How model-family structure becomes explicit compiler input rather than hidden runtime convention.
K-quants deep dive
How Q4_K, Q5_K, Q6_K, Q8_K, and mixed dot products move through real CPU kernels.
Prefill versus decode
Why one model becomes two CPU workloads with different matrix shapes, memory traffic, and execution plans.
Bounded engineering engagements
We take on well-scoped problems where a model, runtime, control system, or Linux CPU path is unsupported, slow, numerically wrong, or difficult to audit.
CPU AI investigation
Feasibility, performance, and first-divergence analysis for CPU inference.
Numerical parity
Runtime-versus-reference divergence localization, operation by operation.
C and Linux optimization
Kernel, memory-path, portability, and profiling work on real hardware.
Controls and simulation
Robotics mathematics, deterministic simulation, and control-system prototyping.
Antshiv Robotics builds intelligent systems we can understand all the way down, from mathematical kernels and CPU runtimes to simulation, flight control, and autonomous machines.