RZ and RA Ready Partner
Renesas
Overview
Our work as an early RZ and RA Ready Partner is a validation of the deep capabilities in AI and Machine Learning, and more specifically our proven track record of implementing highly optimized AI applications on Renesas devices for applications ranging from video, audio and sensor data processing.
Highlights
Neurealm launches AI applications on RA8D1 MCU
The graphics-enabled MCU is the first ARM® Cortex®-M85 core based MCU with Helium, for AI/ML tasks
Our Offerings
We bring deep engineering expertise across Renesas RZ/V MPUs and RA8 MCUs, enabling end-to-end AI, vision, and robotics solutions at the edge.
Embedded Software
- Porting of RTOS/OS and drivers (Ex: Linux, FreeRTOS)
- Custom board bring-up
- Sensor (camera, IMU etc) integration
- Multimedia pipeline optimization (ex: gStreamer)
- Migration of middleware & applications
Perception AI
- Design & development of CV algorithms
- Sensor Fusion (camera, lidar, radar)
- Development of custom Perception AI applications
- Porting & optimizing algorithms/models
- Migration of AI applications
Robotics
- Development & Integration of Navigation Stack
- SLAM (2D & 3D sensors)
- Path Planning
- Rich experience in ROS / ROS 2 environment
- Development of Robotics Software application from Sensor to Actuation
Deep AI Expertise Across
Renesas Products
AI pretrained library
The pre-trained AI library is a collection of Vision AI based software applications available to all. The library and sample applications are a good entry point for beginners to start work with Renesas RZ/V and RA8 series.
- Easy to Deploy — Pretrained AI libraries are readily available sample applications which can be run on the device
- Open Source — The source code, the DNN model and the datasets used for the model training are all open source. Open source ensures wider community support
- Free to Use — The executables, models and source code are all available on GitHub free of cost
DRP-AI Translator Mounting Flow
Applications Build
Smart Vacuum Cleaner Robot
Smart Vacuum Cleaner Robot with RA6M5 and RZ/V2H using ROS2 and micro-ROS
- Base Controller:
- RA6M5 for motor control
- Runs micro-ROS node for motor commands, IMU, and Wheel Odometry
- Processing and Perception:
- RZ/V2H manages 2D LiDAR, camera, and all other software modules for state estimation, navigation and control
- Hardware Accelerated Visual SLAM with DRP-AI and OpenCV for precise localization
- Intelligent Navigation:
- ROS2 Navigation Stack for full coverage
- Indoor AMR mode with path planning and dynamic obstacle avoidance
Human Head Counter
Using RENESAS RA8D1, the first ARM® Cortex®-M85 core–based MCU with Helium™
KEY FEATURES
- Suitable for real time detection on low-power, low-cost edge devices
- Ultra lightweight deep learning models with good accuracy
- Leverages Arm Helium™ technology for high performance
- Able to detect heads from various angles
TARGET MARKETS & APPLICATIONS
- Elevator passenger counting, conference room usage monitoring
- Occupancy-based Lighting and AC control
- Congestion detection in public transport, stations and malls
Other Applications
Driver Monitoring System
• Supported Hardware: RZ/V2L and RZ/V2H.
• Model: TINY/YOLOv3+DeepPose;
• Dataset: WIDERFACE/WFLW
Attire Detection
• Supported Hardware: RZ/V2L, RZ/V2H and RA8D1.
• Model: YOLOv3;
• Dataset: Safety Helmet detection datasets
Animal Detection
• Supported Hardware: RZ/V2L and RZ/V2H.
• Model: YOLOv3;
• Dataset: COCO + Animals Detection Images Dataset
Access Control
• Supported Hardware: RZ/V2L and RZ/V2H.
• Model: Tiny/YOLOv2;
• Dataset: COCO
Office Room Usage Monitor
• Supported Hardware: RZ/V2L and RZ/V2H.
• Model: YOLOv3;
• Dataset: Various datasets
Human Gaze Detection
• Supported Hardware: RZ/V2L and RZ/V2H.
• Model: ResNet-18 + Tiny YOLOv2;
• Dataset: ETH-XGaze + WIDERFACE
Elderly Fall Detection
• Supported Hardware: RZ/V2L.
• Model: TinyYOLOv2 + HRNet;
• Dataset: PASCAL VOC + COCO
Demo Videos
Robot Vacuum Cleaner Demo
Head counting demo on Renesas RZV2L
Safety helmet and vest detection sample application on Renesas RZV2L