We develop custom robotics applications
Focused on HiWonder devices and the NVIDIA Jetson platform for edge AI, autonomy, and real-world robotics software.







Focused on HiWonder devices and the NVIDIA Jetson platform for edge AI, autonomy, and real-world robotics software.






We build end-to-end robotics software—from ROS/ROS 2 architecture and perception to Jetson acceleration, hardware integration, and deployment— so platforms can evolve from demos into dependable systems.
Well-structured ROS/ROS 2 systems: nodes, launch files, parameters, TF/TF2, topics/services/actions, lifecycle patterns, and clean package layouts that scale from prototype to product.
Camera and sensor pipelines for detection, tracking, and scene understanding—designed for low-latency, on-device inference with practical considerations like calibration, synchronization, and lighting variability.
Mapping and localization that supports real environments: SLAM integration, map handling, tuning, and repeatable localization workflows for indoor spaces and changing layouts.
Navigation behavior built around safety and reliability: costmaps, obstacle avoidance, waypoint routing, follow/escort behaviors, and kinematics-aware control tuned to the robot’s platform and environment.
Jetson-first performance work: TensorRT acceleration, CUDA-friendly pipelines, DeepStream/GStreamer video paths, power/thermal profiling, and real-time tuning to keep FPS stable and latency low.
Integration across the stack: motor controllers, encoders, servos, IMU/depth sensors, serial/CAN/I2C/SPI interfaces, plus fault handling and recovery so the robot behaves predictably in the field.
Interfaces that make robotics usable: voice control, basic conversational flows, operator UI, and behavior design focused on clarity, safety, and real-world day-to-day operation.
Product-ready foundations: logging/telemetry, diagnostics, secure connectivity, remote updates (OTA), testing practices, and maintainable architecture so robotics software stays stable over time.
Grandma Monitor is a real home-safety robotics project inspired by a simple fear: if someone falls at home, will anyone know in time? We’re building an AI-enabled robotics application that helps a Jetson-powered platform stay present, understand the home environment, and escalate quickly when something is wrong—while remaining respectful, unobtrusive, and configurable for each household.
Safety & urgency
Detect critical events, reduce false alarms, and trigger the right response sequence fast.
Always-ready operation
Docking/charging readiness, stable runtime states, and predictable recovery when conditions change.
Family peace of mind
Optional secure sharing—streaming, summaries, and alerts that are permissioned and easy to manage.
Project Status
Active DevelopmentBuilt on the same professional stack we deliver to clients: ROS/ROS 2 architecture, Jetson acceleration, sensor integration, and deployable reliability practices.
Autonomy: mapping, localization, navigation, and safe follow behaviors.
Perception: presence, activity, posture signals, and incident detection.
Readiness: charging workflows, watchdogs, logging, and recoverable state design.
Connectivity: optional streaming, notifications, and integrations with sensors/IoT.
Built for real homes, not lab conditions
Emphasis on safety, consistency, and clear escalation—so the system is dependable when it matters.
Grandma Monitor is built around a real story and a real need: staying aware, noticing emergencies quickly, and helping families respond with confidence—while keeping privacy and control in the hands of the people who live there.
Planned capabilities
Roadmap01Always-on readiness
Docking/charging behaviors and power-aware scheduling so the robot stays ready to observe and respond without constant manual attention.
02Home understanding
Mapping and localization to understand the home layout and move efficiently through real spaces—rooms, hallways, thresholds, and obstacles.
03Safe follow + respectful waiting
Human-centered motion that follows at a safe distance, avoids interfering with walking, and can wait intelligently when routines take longer than expected.
04Wellness + emotion signals
Optional cues from expressions and activity patterns to help identify loneliness or distress, paired with gentle engagement behaviors.
05Health device integration
Bluetooth integrations (such as heart-rate monitors) and other sensors to provide additional context for wellness tracking and alerting.
06Family connection
Secure streaming and event summaries so family can check in with confidence, review important moments, and stay connected.
07Fall detection + emergency escalation
Fast incident detection with clear escalation paths—alarms, notifications, and emergency workflows designed to be reliable and configurable.
08Home security + smart home
Intruder detection, alarms, and integration with common IoT devices (like lights and sensors), plus basic health checks for connected devices.
09Daily life recap
Optional daily summaries that highlight meaningful moments—designed to be user-controlled and privacy-aware.
Edge AI
Real-time inference designed for low latency and reliability on Jetson.
Autonomy
Practical mapping and navigation behaviors for real indoor spaces.
Reliability
Watchdogs, logging, safe states, and recovery paths built in from day one.