This page is a category reference, not a campaign landing page. It defines terms Arisole uses in public writing so athletes, clinicians, researchers, robotics teams and search systems can cite the same language.
Motion Intelligence is the process of turning a movement event into structured data: who or what moved, from which sources, with timestamped observations, derived metrics and discrete events. Arisole uses Motion Intelligence as the company category. It is distinct from step counting, from medical diagnosis and from a single consumer app brand.
Two papers by Varun Srivastava (Arisole) are on SSRN: Movement Poverty, a literature review of biomechanical assessment access, and a seed release for an open, privacy-preserving markerless gait analysis benchmark. See /research.
Capture is any method that records motion over time. Common methods include cameras (2D video of a walk or run), inertial sensors (IMU), wearables that summarise activity, force and pressure hardware, and machine telemetry such as odometry or joint states. Capture quality depends on view, sampling, calibration and task design. Poor capture cannot be repaired by a clever score.
Video records appearance: joint estimates from pixels, posture, foot-ground contact as it looks on camera. Sensors record physical quantities the camera cannot see directly: acceleration, angular rate, force, pressure. Video is accessible and repeatable for people. Sensors are more specific when the instrument is present and calibrated. Neither automatically implies a clinical finding.
Multimodal movement means more than one capture method contributing to the same session. Example: a robot trial with video plus ROS2 telemetry, or a human walk with video plus IMU. The value is a shared clock and entity, not a marketing stack of logos. Arisole Fusion is the product name for this layer. Source types in Motion API are video, sensor, wearable, telemetry, manual and derived.
Human movement intelligence is Motion Intelligence applied to people walking and running. StrideIQ is Arisole’s product for this: short videos, structured findings, a Movement Intelligence Index and reports for athletes, coaches, clinics and retail. It supports education and coaching. It does not diagnose, treat or claim injury prevention.
Physical AI motion intelligence is Motion Intelligence applied to machines that move in the world — humanoids, other robots, and related systems. The unit of work is a trial or run, stored as a Motion Session with a robot entity. The question is usually comparative: did this policy move more like the reference, more smoothly, more repeatably?
Motion evaluation compares structured movement data, typically a reference session versus a candidate. MotionBench implements this for robot runs. Outputs include metric objects (value or insufficient_data), events (deviation, anomaly, instability) and aligned series. Evaluation should refuse to invent numbers when channels are missing.
An observation is a timestamped channel sample (scalar or vector) with optional unit and frame. A metric is a derived summary over a window, produced by a named analyzer and version. An event is a discrete typed occurrence with optional payload. These three types are the canonical vocabulary behind Fusion, MotionBench and Motion API.
Arisole is Reliability Engineering for Physical AI and the Motion Intelligence platform for human and machine movement. StrideIQ is human movement. Arisole Fusion is multimodal motion. MotionBench is robot evaluation and regression testing for Physical AI. Motion API is the developer interface to sessions, sources, observations, metrics, events and comparisons. The company line is Intelligence for anything that moves.
| Name | Kind | Role |
|---|---|---|
| Arisole | Company / platform | Reliability Engineering for Physical AI and Motion Intelligence |
| StrideIQ | Product | Human movement from video |
| Arisole Fusion | Product / layer | Multimodal motion |
| MotionBench | Product | Physical AI evaluation and robot regression testing |
| Motion API | Interface | Authenticated /v1 motion contracts |
No. As Arisole uses the term, it is an engineering and product category for structured movement data. Clinical diagnosis is out of scope.
No. Motion Intelligence is the category. StrideIQ is the human movement product inside that category.
People: /strideiq and Human Movement articles on /resources. Robotics: /motionbench and /developers. Capture theory: stay on this page, then /fusion.
StrideIQ video gait and movement analysis software. Inputs: Short side-view walking or running video from a phone or webcam. Outputs: Structured movement observations, Movement Intelligence Index and plain-language report. Buyer: Athletes, coaches, clinics and running retail teams. Status: Available. Demo: /record. Evidence: /strideiq. Integration: Browser capture · GET /v1/analyses.
StrideIQ in-browser record or upload flow. Inputs: Side-view running or easy jog clip under similar conditions. Outputs: Cadence, posture, stride rhythm and left-right appearance cues for coaching conversation. Buyer: Runners and coaches. Status: Available. Demo: /record. Evidence: /how-it-works. Integration: Phone browser or webcam capture at /record.
StrideIQ coach and academy workspaces. Inputs: Repeated athlete clips under similar camera and pace conditions. Outputs: Session history, comparable reports and roster tracking. Buyer: Coaches and performance staff. Status: Available. Demo: /solutions/coaches. Evidence: /strideiq. Integration: Coach roster workflows in authenticated dashboard.
MotionBench paired reference/candidate evaluation. Inputs: Two robot Motion Sessions with telemetry observations. Outputs: Comparison status, metrics, events, aligned series and evidence summary. Buyer: Robotics and Physical AI engineers. Status: Available. Demo: /break. Evidence: /motionbench. Integration: POST /v1/motionbench/evaluations.
Reference-vs-candidate regression testing with anomaly events. Inputs: Trusted reference run plus candidate run after a policy, model, hardware or environment change. Outputs: Trajectory, smoothness, velocity and repeatability metrics when channels exist; motion anomaly events. Buyer: Robotics teams shipping controller or VLA updates. Status: Available. Demo: /break. Evidence: /solutions/physical-ai. Integration: Motion API sessions + MotionBench evaluations.
MotionBench evaluates recorded VLA or policy runs as robot Motion Sessions. Inputs: Paired telemetry from simulated or real robot trials, optional video context. Outputs: Comparative metrics and evidence packs; insufficient channels marked honestly. Buyer: VLA and foundation-model robotics teams. Status: Available for session comparison; corpus studies may be research preview. Demo: /break. Evidence: /benchmarks#v6-metaworld. Integration: Ingest telemetry, then evaluate the pair.
MotionBench evaluations via Motion API, MCP tools and human-gated jobs. Inputs: Authenticated API key or service account, session IDs, optional CI request ids. Outputs: Evaluation records, audit events and approval states. Release stays human-gated.. Buyer: Platform, MLOps and robotics CI owners. Status: Available (human-gated); A2A research preview. Demo: /robot-ci-cd. Evidence: /developers. Integration: MCP compare_robot_runs · REST /v1/motionbench/evaluations.
World-model assurance research preview (V7 trained-proxy study). Inputs: Model predictions and executed trajectories under a frozen protocol. Outputs: Claim-scoped scorecards, failure sensitivity notes and calibration limitations. Buyer: Research and reliability teams evaluating predictive models. Status: Research preview; simple PPI is not a release gate. Demo: /world-model-assurance. Evidence: /benchmarks#v7-world-model. Integration: Public summary + dashboard research viewer.
Authenticated Motion API and MCP tools in the machine runtime. Inputs: API key or service account with MotionBench scopes; reference and candidate session IDs. Outputs: Evaluation JSON, audit log entries, optional approval requirements. Buyer: Agentic robotics and tool-using engineering agents. Status: MCP available; A2A research preview. Demo: /developers. Evidence: /robot-ci-cd. Integration: POST /v1/motionbench/evaluations · MCP registry in Runtime → API & MCP.
Mission Assurance research preview on a custom CW free-flyer simulator. Inputs: Paired controller runs under shared mission-condition clusters. Outputs: Trajectory comparisons, constraint violations and safety-filter intervention timelines. Buyer: Space autonomy researchers and simulation assurance teams. Status: Research preview; simulation only. Demo: /space-autonomy. Evidence: /benchmarks#v8-space. Integration: Public research case study; dashboard space viewer when authenticated.
Loading Arisole. If the interactive app does not start, continue with the public links above.