Technology
From motion signals to mechanical insight
Position tells us where a mechanism is. Velocity tells us how it moves. Motor current provides information related to the effort required. Studying these signals together can help characterize operating behavior and reveal changes that deserve attention.
Position
Where the mechanism is within its range of motion.
Velocity
How fast it moves through that range.
Current
Information related to the effort the motor applies.
3D position–velocity–current modeling
The existing 2ndSight analytics foundation models relationships between position, velocity, and current during operation. A learned reference provides a basis for assessing changes in machine behavior as the machine repeats its work.
The goal is to help engineers identify where behavior changes within the motion range and investigate whether those changes reflect operating conditions or mechanical issues.
Friction and inertia estimation
2ndSightAI builds on friction estimation methods and is developing the Q-Connect workflow for estimating friction and inertia from motion data. The approach uses velocity, acceleration, and current. Where acceleration is unavailable, it can be estimated from changes in velocity over time.
Trends in these estimates may help engineers investigate changes in resistance or load. Useful results depend on suitable motion, signal quality, and validation for the application.
Physics-informed stream clustering at the edge
We plan to use micro-clustering and stream clustering to analyze incoming position, velocity, and current (PVC) data on Q-Connect devices. Compact cluster summaries will be updated as data arrives, helping characterize recurring motion patterns and identify changes near the machine.
We are evaluating the Q-Connect implementation for data quality, memory use, processing load, and behavior under changing operating conditions. The final approach will be selected and validated through testing on representative motion data.
Selected cluster summaries and analysis results are intended for cloud visualization and further comparison. Comparisons across axes and machines will require consistent features and operating context; local cluster labels alone are not directly comparable.
Status labels reflect our current understanding and will be updated as capabilities are validated.
From the machine to the dashboard
How the platform is designed to work
-
Servo drive
The planned workflow reads position, velocity, and current through supported service or configuration interfaces.
RS-232 · RS-485 · USB · Ethernet -
Q-Connect edge gateway
Planned edge analytics and micro-clustering of streaming PVC data.
In development -
Cloud platform
The planned cloud workflow receives selected analytics, cluster summaries, and device status from the edge.
In development AWS · MQTT over TLS -
Analytics dashboard
Organize devices, inspect models, review health trends, and investigate changes.
In development
Planned data path. Q-Connect is designed to use service or configuration interfaces; it does not tap the EtherCAT or CANopen motion-control bus. Each drive model requires integration and validation. Available signals, sampling rate, and timing determine which analyses are practical.
Connect
Q-Connect accesses available signals through supported service or configuration ports, such as RS-232, RS-485, USB, or Ethernet. The planned data path does not tap the EtherCAT or CANopen control bus.
Analyze
Edge software processes motion data close to the machine. The available analytics depend on signal access, timing, and data quality.
Review
The cloud platform is being developed to organize connected devices and present models, health trends, and analysis results for engineering review.
Development focus
Our next priorities are validating third-party PVC data access, evaluating micro-clustering and stream clustering of PVC data on Q-Connect, and connecting edge results to the cloud dashboard. The implementation and supported drive models will be confirmed through testing.
Help build the next stage
We are bringing together motion engineering, machine learning, and cloud software to develop a practical motion analytics platform. Join the founding team or help shape the product through a pilot discussion.