Autonomous driving vehicle control by edge computer network having very short response time has been proposed. Each vehicle has an agent program on the edge and automatically moved to adjacent edge computer following to the vehicle movement. On the edge computer, the agent program is processing with other vehicles' agents in a cyber network. We employ triple redundancy and majority rule to achieve high-reliability and less than 10 ms control latency try to be guaranteed. In addition, each vehicle has an IoT sensor including fine-GPS, so all precious position and speed of the vehicle can be monitored. We constructed the autonomous driving experimental course, the autonomous driving vehicle, and the edge computer system with the center cloud and tested in the campus test bed. According to the vehicle movement, network orchestrator setup a new optical path automatically and send the agent program to the adjacent target edge computer. This orchestration function is newly proposed application triggered dynamic optical network. In this presentation, I will show the detailed experimental results. This architecture and experimental results can be applied to the future smart and connected community.