Plant Bot, robotic learning feedback


Artist David Bowen’s installation Plant Bot experiments with the time and ways of interdependence between a plant and a system consisting of a robotic arm, a video camera and a computer. The video camera, mounted on the robotic arm, observes the health of the plant using an object-detection model by artificial intelligence running on the computer to which it is connected. The robot “searches” for the size, location and type of leaf, categorising each as either “happy,” or “dead.” When 1000 of these images are collected, the computer uses these, along with their annotations, to lay out the training guidelines for an artificial intelligence model. Contextually, the system uses the ratio of healthy to dead leaves to determine in real time what care to give the plant by watering it and exposing it more or less to light. If the plant does not appear healthy to the computer, the system will attempt to help it by applying the behaviour it deems appropriate with respect to the collected image history. The life expectancy of the plant is thus, therefore, directly related to the system’s ability to learn and become more proficient. This living sculpture revolves around a caring vegetable machine, called upon to learn according to its statistical-deductive possibilities. The forced symbiosis goes through a series of successes and mistakes, and although there is no certainty of achieving the desired results, it stages an essential and metaphorical relationship that makes symbolic gesture its strong point. Benedetta Sabatini


David Bowen – Plant Bot