Dikole, Realeboga GFaniso-Mnyaka, ZimbiniSekopa, Teboho L2025-05-022025-05-022024-12979-8-3315-1815-8979-8-3315-1814-1979-8-3315-1816-5DOI: 10.1109/ICCR64365.2024.10927581http://hdl.handle.net/10204/14229Reinforcement learning-based robotic manipulation has risen to prominence recently due to the rise of frameworks such as Open AI Gym. Although there has been much success in reinforcement learning applied to manipulators such as the Kuka iiwa, Franka Emika Panda, UR5 robots, there has been little to no exploration of the application of reinforcement learning on ABB robotic manipulators. This paper presents a perspective on integrating vision-based reinforcement learning and ABB robots with Open AI Gym environments. We focus mainly on the pick-and-place and push environments using proximal policy optimisation due to its simplicity and ease of memory. Our results affirm the possibility of vision-based reinforcement learning with ABBrobots with the best training performance yielding a success rate of close to sixty-five percent.AbstractenReinforcement learningRoboticsManipulationConference Presentationn/a