Haptic Object Recognition

在科学文献中有一个共识，即开发更像人类的触觉感知能力是设计新一代智能机器人的不可或缺的要求。为了应对这一挑战，该提案将研究和开发新的、更有效的方法，用于实时机器人触觉对象识别，这有助于提高机器人的触觉灵敏度和敏锐度，使其更接近人类水平。 所提出的研究的主要目标是开发新的计算智能算法的实时机器人触觉感知触摸物体表面拓扑结构和基于模型的触觉只有3D物体识别。更具体地说，将开发新的算法，用于整合触觉传感器图像和动觉传感器数据，这些数据是通过机器人对3D物体的触觉探索收集的。任务驱动的主动传感方法将进行调查，以允许一个固有的有限的触觉传感能力的有效利用。将开发用于机器人操作器的位置和力并发控制的新的智能控制算法。新的计算智能方法，基于模型的触觉识别的3D对象，使用从触觉图像恢复的拓扑特征也将开发。 拟议的研究将提供急需的科学知识和技术创新，以帮助提高机器人的触觉灵敏度和敏锐度，使其更接近人类的范围，提供实用的解决方案，最大限度地减少触觉探测工作，同时最大限度地提高物体识别效率，比今天更广泛的物体。所有这些都将允许开发新一代智能机器人，能够在能见度差或不存在的条件下执行复杂的操作操作，例如水下，危险或高风险的操作环境-例如核电厂的维护和爆炸装置的处置，或远程医疗-其中触觉至关重要。

There is consensus in the scientific literature that the development of a more human-like haptic perception capability is an indispensable requirement for the design of a new generation of intelligent robots for advanced applications. Rising up to this challenge, this proposal will investigate and develop new and more efficient methods for real-time robotic haptic object recognition that can help raise the level of tactile sensitivity and acuity of robots closer to the human level. The main objective of the proposed research is to develop new computational-intelligence algorithms for real-time robotic haptic perception of touched object-surface topology and for model-based haptic-only 3D object recognition. More specifically, new algorithms will be developed for the integration of tactile sensor images and kinaesthetic sensor data collected through robotic haptic exploration of 3D objects. Task-driven active-sensing methods will be investigated to allow for an efficient use of inherently limited tactile sensing capabilities. New intelligent control algorithms for concurrent position & force control of the robot manipulator will be developed. Novel computational intelligence methods for model-based haptic recognition of 3D objects using topological features recovered from haptic images will be also developed. The proposed research will contribute much needed scientific knowledge and technological innovation to help raise the tactile sensitivity and acuity of robots closer to the human range by providing practical solutions that will minimize the haptic probing effort while maximizing the object recognition efficiency for a wider set of objects than possible today. All these will allow for the development of a new generation of intelligent robots able to perform complex manipulation operations under poor or nonexistent visibility conditions, such as underwater, in hazardous or high-risk operational environments - as for instance maintenance in nuclear plants and disposal of explosive devices, or in telemedicine - where touch feeling is of paramount importance.