Physical automation, made comprehensible

Reliable robotic orchestration
for work previously impractical to automate.

Comprehensible is a control and orchestration system for physical workflows across robots, sensors, AI models and business systems — with safety, monitoring and recovery built in.

Better perception and decision-making are opening automation to changing, lower-volume work — from smaller manufacturers and creative studios to construction sites and farms. Comprehensible makes these systems practical to set up and adapt without a dedicated team of robotics specialists.

Explore the product

01 / THE PRODUCT

Automation no longer has to mean a rigid, single-purpose production line.

From device drivers and motion control to perception and workflow logic, Comprehensible connects the full stack into adaptable workflows. Smaller manufacturers, workshops, wineries, breweries and creative teams can design, test and operate changing jobs without creating a custom integration each time.

+

Compose the whole workflow

Connect live sensor data, spatial understanding, decisions and robot actions.

+

Keep the real system in view

Observe connected hardware while designing, testing and adjusting the job.

+

Build the interface for the task

Combine the controls, status, simulation and timing that operators actually need.

REAL-WORLD EXAMPLE

Bringing robotics into a product filming studio

Comprehensible orchestrator for a real product filming studio, connecting live LiDAR data, scene recognition, a user prompt, camera trajectory and robot execution while showing current hardware state
01 Design the workflow while observing the current state of connected hardware
Comprehensible interface showing a robot arm, a natural-language motion assistant, the generated motion program and its timeline
02 Natural-language robot motion, visualized and editable
A task-specific interface created in Comprehensible, combining a multi-device simulation, live status and a shared motion timeline
03 A task-specific control interface built in the platform
CONNECTED HARDWARE / REAL SYSTEMS

The interface is connected to real machines and real sensor data.

A real collaborative robot used with the Comprehensible motion interface
01 Collaborative robot
A LiDAR point cloud showing a real workspace and detected spatial regions
02 LiDAR perception and spatial regions
A real motorized camera rig orchestrated through a shared timeline
03 Multi-axis camera rig
A Tandee robotic pallet-nailing cell operating an HS head over an XL worktable
04 Robotic pallet nailing

02 / THE FOUNDATION

Intelligence makes new automation possible. Reliability and safety make it viable in practice.

Comprehensible’s reliability and safety engine sits at the foundation of the orchestration layer. Every input, action and outcome has defined meaning, allowing the platform to check the complete workflow against its safety rules, live sensor readings, and the actual positions and status of connected equipment — before anything moves and throughout testing. In production, the same foundation continuously monitors events across space and time, detects anomalies in real time, and triggers a defined response when conditions change.

SPECIFICATION LAYER

To make these requirements explicit and machine-checkable, we designed our own domain-specific language for workflow and safety specifications. Specifications can be written directly or generated with an LLM, then validated before they can affect physical equipment.

BEFORE DEPLOYMENT / VALIDATION

Understand how the machine behaves before trusting it with the job.

Time-series joint diagnostics comparing commanded and measured position, velocity and torque
01 Compare commanded and measured position, velocity and torque over time.
Robot joint friction profile plotted from sensed velocity and torque
02 Expose friction, load and limit behaviour from measured motion.
Friction model fitted to measured robot joint telemetry
03 Fit machine-specific friction models from measured telemetry.
A machine-checkable specification linking commanded and measured roll and pitch signals
04 Express causal requirements as machine-checkable specifications — written directly or generated with an LLM.
DURING OPERATION / MONITORING

Keep the robots, their surroundings and their history in view.

A technical workstation showing robot simulation, live telemetry and diagnostic tools
05 Inspect robot state, telemetry and diagnostic tools together during development and testing.
LiDAR point cloud and camera view used to monitor people, objects and spatial regions
06 Fuse LiDAR and camera data to monitor people, objects and safety regions in the same physical scene.
A time-aware spatial map of changing traffic signal states with data and monitoring logic
07 Replay and inspect events across space and time — here, changing signal states across a road network.

Build with us

Physical automation,
made comprehensible.

Bring us the process you want to automate. We’re working with early teams to turn high-value, human–robot use cases into reliable operations.

admin@swarmis.us