Technology
Radio Frequency & Machine Vision Technologies
Position Imaging brings 20+ years of R&D in radio frequency and machine vision, backing one of the industry’s deepest indoor positioning and computer vision patent portfolios.
Our Foundations
Position Imaging’s patent portfolio is independently cited by 24 of the world’s leading technology, industrial, and logistics companies, spanning Consumer Electronics & Computing, Enterprise & Networking, Logistics & Commerce, and Automotive & Industrial, an independent signal of the reach and relevance of our intellectual property.
Computer Vision Logistics
Not everything worth tracking can carry an RF tag. Packages that go missing between the dock and the door, retail items lost to shrinkage, minutes workers waste searching shelves instead of restocking – these costs pile up because tagging every object isn’t practical. Vision fills that gap: cameras, depth sensors, weight sensors, and machine learning working together to confirm exactly what was placed, moved, or removed.
A shelf-mounted camera with a field of view fixed on the shelf below it can detect, from image comparisons alone, whether an object has been placed, moved, or removed – and a co-located directional light source (laser or LED) can then guide a person’s hand to exactly the right spot, whether that’s a drop-off location or the item they’ve come to retrieve.
How We Did It: 24+ patents, granted 2018-2026, covering the hardware, software, and machine learning behind our smart-room and package-intelligence product lines.
Weight sensors cross-check what the camera sees, flagging a mismatch between the expected and actual handling of an item, catching errors or improper removal that vision alone might miss. Continuous video monitoring adds a further layer, detecting rule-violating handling (a thrown package, the wrong item taken) and triggering real-time alerts.
The camera, depth sensor, and directional light live together in a single, rail-mountable module, engineered for fast retrofit into existing shelving, with a companion calibration technique that aims the light precisely at a target in one motion, without the trial-and-error most laser-guidance systems require.
A deep neural network identifies objects and reads their labels or barcodes automatically. When it fails to recognize something, the system asks a person once, then retains itself on that image, compounding accuracy over time across every deployed unit without manual retraining cycles.
The same underlying technology powers full package-management deployments: cameras and beam-break sensors (for thin items like envelopes that vision alone tends to miss) register drop-offs and pickups, notify recipients, and guide them to the correct location – deployable as a wall-mounted system or a free-standing kiosk with an integrated scanner.
Sensor Fusion & Continuous Tracking
A raw RF position fix is just one data point; a single fix is easy. What determines whether a deployment survives beyond day one is a tracking system that holds up in a real building, with dead zones, signal collisions, server outages, and forklifts moving at speed. Turning that raw signal into something stable, continuous, and scalable enough to run a warehouse or guide a vehicle is a separate, equally hard problem.
Position Imaging’s tracking devices pair the RF position engine with onboard inertial sensors (accelerometers, gyroscopes, magnetometers), blended through Kalman filtering. The result is a system that keeps tracking smoothly through momentary signal loss or multipath interference, and that can power down its radio when a tag is stationary to extend battery life and waking automatically the instant motion resumes.
The same positioning math extends to autonomous and guided vehicles – a vehicle-mounted receiver array determines its position relative to a tracked object or beacon, optionally combined with an absolute reference (e.g., GPS) carried in the same signal – and to broad, non-indoor geographic tracking for augmented and virtual reality applications operating beyond a single room.
Rather than relying on a central controller to calculate every position, a mesh of self-configuring nodes can broadcast their own identity and location, letting the system scale, self-heal, and resist RF interference without a single point of failure: a meaningful advantage over conventional real-time locating systems (RTLS) architectures.
A mobile barcode or RFID scanner’s own tracked location can be used to tag the location of whatever it just scanned, letting a handheld device serve as a mobile position source for inventory and asset placement, without requiring every tracked item to carry its own RF tag.
76 Granted Patents. 61 Pending Applications.
IP Portfolio
Position Imaging’s patent portfolio spans all seven technology areas above — representing over two decades of proprietary innovation in RF positioning, computer vision, and AI. If you’re researching licensing opportunities or evaluating PI’s IP for your industry, the patent portfolio is where to start.
