JAKA has launched a new product in China: the JAKA Lumi. The name is derived from ‘Lux’ (Latin for ‘light’) and ‘mini’ (from the JAKA Mini series), symbolizing the spark of wisdom light through agility, flexibility, and openness. It’s an embodied intelligence platform that offers an all-in-one training solution, now available in China at an exceptionally cost-effective price.

The global embodied AI market has great potential, but traditional solutions face several challenges to widespread adoption.

Pain Point 1: Delayed Deployment and High Costs.

Developers typically procure components like mobile bases, cameras, and robotic arms from fragmented channels. This low-volume purchasing leads to customization difficulties, as well as longer delivery and integration cycles. With high single-unit costs and few ready-to-use solutions, scaling research and training becomes challenging.

JAKA Solution:

JAKA Lumi uses JAKA's integrated drive-control joints, controllers, and robotic arms, all connected by a unified control system that enables setup in just 5 minutes. Its open APIs and modular design make it easy to expand and develop, significantly reducing hardware costs compared to traditional solutions. Currently available in China at an industry-leading low price, JAKA Lumi offers developers a highly cost-effective platform for robot training and R&D. International sales are pending certification.

Hardware: 6-DOF robotic arm with a 2 kg payload, ±160° waist rotation, ±180° neck rotation, 0-400 mm vertical workspace, and compatibility with various JAKA+ ecosystem end-effectors, including 3-finger and 5-finger dexterous hands and flexible soft grippers.

Perception: A triple perception system featuring a chassis-mounted laser radar, 3D vision, and ultrasonic obstacle avoidance enables autonomous mapping and positioning. This ensures the JAKA Lumi platform can seamlessly navigate and avoid obstacles in high-traffic environments.

Pain Point 2: Lack of Integrated Embodied Intelligence Systems.

After purchasing various hardware components, developers who wish to enable data collection, large model training, and deployment still need to independently call APIs, develop corresponding software and algorithms, and deal with high integration complexity, making development time-consuming and labor-intensive.

JAKA Solution:

Open Ecosystem: JAKA Lumi provides an efficient and comprehensive training toolchain that seamlessly integrates with large model platforms and supports mainstream simulation platforms such as MuJoCo and NVIDIA Isaac Sim.

Accessibility: Graphic programming interface and standardized APIs enable basic functions without the need for deep learning expertise.

Performance:

1. JAKA Robohub controller with a 12th Gen Intel i5 CPU (optional RTX 3050 GPU) for enhanced processing power.

2. Localized model inference reduces dependence on cloud computing.

3. Supports EtherCAT bus communication and parallel computation of multimodal algorithms for smooth and synchronized operation.

With training and development, JAKA Lumi can expand its application in areas such as restaurants, hotels, homes, and elderly care.

1. In commercial settings like food delivery, hotel services, and retail, JAKA Lumi can autonomously plan paths, deliver items, and serve food.

2. Its voice module enables it to understand commands and deliver items like towels or chargers, enhancing customer satisfaction.

3. The lifting system adapts to different shelf heights, and with visual recognition, it supports 24-hour unmanned inventory management.

4. In home and elderly care, JAKA Lumi's open APIs allow quick development of household robot prototypes, assisting elderly users, calling emergency contacts, and integrating features like visual detection and voice reminders.

JAKA Lumi proves that high performance doesn't have to come at a high cost. Through continuous improvement in integrated control, multimodal perception, and an open ecosystem, businesses, researchers, and developers can work together to define the boundaries of intelligence on common ground.