From pixels to position.
Perception software that turns image and inertial data into useful geometric information.
A perception stack, not just a camera feed.
VILOTA combines imaging hardware and embedded computation with the software needed to estimate movement, describe nearby space and interpret visual scenes. Each application can use the modules appropriate to its platform and operating conditions.
Core perception capabilities
Explore the information that vision systems can produce from multiple sensor streams.
Visual-inertial odometry (VIO)
Visual feature tracking is combined with inertial measurements to estimate changes in position and orientation over time, helping a platform maintain a local pose estimate even when GNSS is limited.
Visual positioning and map matching
Visual observations can be compared with known scene features or maps to assist localisation. Integrations may combine these estimates with GNSS or other host-system position references.
Stereo depth and point clouds
Calibrated stereo imagery yields disparity and depth information. Depending on configuration, the software can support obstacle awareness, 3D point-cloud generation and occupancy representations.
Object detection and tracking
On-device computer-vision models can identify and follow relevant objects, with model selection and processing rates determined by compute resources and application needs.
TagSLAM and visual landmarks
Recognised markers and revisited features can support tracking, loop closure and localisation workflows in mapped or structured environments.
Low-light visual features
Sensor selection, exposure, gain control and feature extraction are considered together for low-texture or dim scenes. Performance depends on sensor, optics, lighting and software tuning.
How the pipeline fits together
Illustrative software workflow; actual enabled modules vary by product.
Observe
Capture synchronised camera and inertial data, with optics and exposure selected for the mission.
Estimate
Track stable image features, fuse inertial signals and compute pose and depth where configured.
Interpret
Produce navigation-ready outputs such as trajectories, point clouds, occupancy cues or detection tracks.
Integration considerations
Reliable perception depends on more than choosing an algorithm.
Multi-camera geometry
Camera placement, field-of-view overlap, lens selection and rigid mounting influence feature tracking and depth estimation.
Calibration and synchronization
Intrinsics, extrinsics, time alignment and inertial calibration are central to interpreting multiple streams together.
Compute and deployment
Performance is shaped by image resolution, number of streams, CPU/AI acceleration, latency, heat dissipation and power budget.
Explore the hardware behind the software
Perception capabilities are deployed on a matched sensing and compute configuration.
DP180 integrated platform
Stereo depth, VIO and onboard processing in an integrated perception device.
↗Coreboards & compute
Select a platform and interface arrangement around the required workloads.
↗Discuss a vision workload
Share your pose, depth, AI and integration requirements with engineering.
Some functions shown here are software-stack capabilities or development options rather than universally enabled features on every product. Ask VILOTA for the supported configuration.
Have a vision application in mind?
Talk to our team about sensors, compute, software and integration.