For many routine surgeries and procedures, such as liver biopsies, obese patients present an increased challenge. In liver biopsies, it is more difficult to visualize the needle tip in realtime abdominal ultrasound images, particularly when multiple layers of fat and other tissues obscure needle localization, which results in multiple needle passes to obtain an adequate biopsy sample. This work aims to develop a robotic system that can autonomously and robustly follow a biopsy needle regardless of the tissue medium by using photoacoustic imaging. Our system consists of a robotically controlled ultrasound probe that continually visualizes the location of the biopsy needle tip by segmenting photoacoustic signals generated from an optical fiber inside the needle. This system was able to track and remain centered over a needle inserted in fat, muscle, and liver tissue with mean errors of less than 1mm, and it successfully recovered from a perturbation of the ultrasound transducer that caused the needle to exit the image plane. Our results show promise for guiding biopsies in obese patients with this novel system.