AI Model
AI model development and fine-tuning
Select and develop AI models for the required perception and decision-making tasks.
Fine-tune them with company-specific data and reduce their size as needed for the target hardware.
Edge AI
Edge AI
Deploy AI on devices in the field
We develop AI for edge devices such as NVIDIA Jetson and connect it with sensors and business systems to meet your needs.
We also develop and fine-tune compact, fast AI models, targeting a balance of accuracy and processing speed within hardware and connectivity constraints.
Building on prepared data and simulation results, we validate and refine the system on physical devices for use in real-world operations.
3 Development Areas
Three development areas for AI deployment in the field
On-device AI
AI deployment on field devices
Integrate AI models into edge devices such as NVIDIA Jetson and implement the path from data input to inference output.
Optimize operation for processing speed, memory, power consumption, and connectivity constraints.
System Integration
Sensor and business-system integration
Connect sensor data and AI inference results with existing equipment and business systems.
Develop visualization, alerts, and logging so results can be used in daily operations.
Work Type
Examples of work
Requirements and hardware design
Review the workflow, required accuracy, response time, and installation environment to design the model, hardware, and system configuration.
Evaluate devices such as NVIDIA Jetson Orin Nano and Jetson AGX Orin against model size, memory, power, and sensor-interface requirements.
Model development and refinement
Select and develop models suited to the task and fine-tune them with company-specific data where needed.
Evaluate accuracy while reducing model size and refining inference.
On-device implementation
Implement inference software on the target device, connecting preprocessing, inference, and output.
On Jetson, use tools such as NVIDIA JetPack and TensorRT to optimize inference for supported models.
Measure accuracy, processing time, and memory use on the hardware and refine its operation.
Sensor and system connections
Design connections to sensors and existing systems, then implement data acquisition, result transmission, and logging.
Develop monitoring screens and alerts where needed.
Field validation and improvement
Check accuracy, response time, and sustained operation under actual operating conditions.
Identify issues, including connectivity loss and changing inputs, and improve both the model and the system.
EXDREAM R&D / Robot Decision System
Edge development with RDS
Robot Decision Data
Data for learning operational decisions
Our Robot Decision System (RDS) creates decision-learning data from device logs, sensor data, operator histories, and simulation results.
We are developing datasets for edge-based decisions that capture operational states, candidate actions, selected actions, and outcomes.
Robot Decision Model
Lightweight decision models for the edge
We are researching and developing lightweight AI models that evaluate and select the next action for specific companies and tasks.
Through distillation, fine-tuning, and model-size reduction, we aim to achieve low-latency decision-making on target devices such as Jetson.
Robot Decision Harness
Decisions and execution within device constraints
We are researching and developing mechanisms that adjust model selection, action-candidate counts, and other decision settings to task complexity, risk, response-time requirements, and available compute.
We evaluate decision quality, latency, and memory use through simulation and hardware tests to support future edge operation.
DELIVERABLES
Delivery overview
AI models and software for your application
Deliver developed and refined AI models and inference software in agreed formats, together with operating requirements, configuration guidance, and other information needed for use.
A runtime connected to devices and systems
Set up an environment where AI runs on the target hardware and connects to sensor inputs and existing systems.
Include result displays, alerts, and logging as agreed.
Field deployment and operational handover
Deploy the system based on hardware validation results and hand over operating, update, and troubleshooting procedures.
Support further evaluation and improvement within the agreed project scope.
Step
Edge AI development and deployment steps
Define requirements
Review operational needs and conditions to define functions, accuracy, response time, and target hardware.
Develop the model
Use data and simulation results to select, develop, and fine-tune the model.
Deploy on devices
Integrate AI on the target hardware and optimize it while checking accuracy, speed, and memory use.
Connect systems
Connect sensor inputs through to the use of inference results in operational workflows.
Validate and launch
Validate and improve operation in the field, then deploy and hand over to operations.
3 OFFERING TYPES
Three engagement models tailored to your goals, scale, and implementation complexity
Based on your team structure and your project’s scope, scale, and goals, you can choose from the following three engagement models.
Around once a week
1–3 people
Consulting
Consulting support for development project planning and progress
We create development plans for each company’s projects, document requirements, and support development team formation and vendor negotiations.
We also review actual progress and provide support to keep development projects running smoothly.
Around three days a week
1–3 people
Tech Lead
Support in-house development as your development team lead
We join your development team as your tech lead and work alongside you on design and development.
We share practical expertise and provide educational content, including seminars, so your employees can continue development independently after the project ends.
Embedded on site
Up to 5 people, depending on project scale
FDE
Advance your development project with our team embedded on site
EXDREAM’s development team works on site at your company, handling everything from planning and coding to deployment.
For companies without engineers, we serve as your dedicated development team.
For companies with an existing engineering team, we join forces and work together toward successful delivery.
Services, team size, engagement frequency, duration, and responsibilities can be flexibly tailored to each project.
START A PROJECT
Edge AI
Discuss development and deployment
Tell us about your challenges and goals for AI model development and deployment on devices in the field.We support you through consulting, in-house development support, and FDE.