A robotic adaptive door opening method with a combination of force and vision

Abstract The automatic door-opening operation is critical for many tasks. Yet, the relevant systems mostly depend on precise visual location, and the inevitable errors may cause fatal harm to the robotic system. Furthermore, no force-control techniques have been specifically designed for them. This research addresses these issues by proposing a hybrid complementary control strategy. Firstly, a two-stage active visual positioning method from coarse to fine was introduced to locate the door handle efficiently. Secondly, a strengthened force control method was proposed, and the force errors and countermeasures in the door-opening process were systematically analyzed. Thirdly, a versatile door-opening system was designed and implemented. Real-world trials were carried out to verify the effectiveness. As the results revealed, with comparative small contacting force and torque, the entire process from recognition to successfully door-opening operation can be completed in 12 seconds at the fastest. In addition, compared with several state-of-the-art door opening methods, the proposed strategy was faster, with less interactive force and torque, demonstrating the superior performance of the proposed method in complex robotic operations.