This post is also available in:
עברית (Hebrew)
A new generation of autonomous agricultural vehicles is emerging, designed to handle field operations with the precision and adaptability of human workers—without human drivers. At FIRA USA 2025, California-based robotics firm Bonsai Robotics unveiled three vehicles under its Amiga line, all powered by its proprietary vision-based autonomy platform, Bonsai Intelligence.
The lineup includes the Amiga Flex, Amiga Max, and Amiga Trax, each developed to automate a range of tasks such as spraying, towing, weeding, and hauling across varied terrains. Central to all three platforms is Bonsai Intelligence—a system that combines advanced perception, navigation, and real-time decision-making to enable fully autonomous operation in complex outdoor environments.
Unlike traditional GPS-guided systems, Bonsai’s approach relies primarily on computer vision and AI to interpret surroundings and make decisions on the move. Trained on data from more than 2 million square meters of global agricultural operations, the platform adapts to changing light, dust, and weather conditions. The system includes two integrated components: Bonsai Autonomy, the onboard perception and control software, and Bonsai Pilot, a cloud-based interface for remote planning, monitoring, and updates.
The Amiga Flex serves as a compact, modular research and development platform. Comparable in size to a small ATV, it supports a wide range of implements and sensors for experimentation in field robotics. Its three-point lift can raise 315 kgs, carry up to 360 kgs, and tow loads of 725 kgs. The design incorporates open data and power interfaces for easy integration with custom systems, while a swappable battery pack delivers up to eight hours of operation per charge.
According to Interesting Engineering, the Amiga Trax and Amiga Max extend this autonomy to larger, production-ready applications. Trax is optimized for rugged, low-clearance environments such as vineyards and cane fruit cultivation, while Max combines compact size with heavy-duty capability, including a hybrid-electric option for extended work cycles.
By integrating vision-based AI directly into vehicle hardware, these systems demonstrate how autonomous robotics can move beyond research and into daily agricultural use—offering scalable, data-driven automation suited to both small farms and industrial operations.
The project’s site can be found here.
