Robotic fingers with multidirectional tactile perception are of great importance for the robotic exploration of complex unknown space, especially in environments in which visualization is not possible. Unfortunately, most existing tactile sensors for robotic fingers cannot detect multidirectional forces, which greatly limits their potential for further development in navigating complex environments. Here, we demonstrate a soft magnetoelectric finger (SMF) that can achieve self-generated-signal and multidirectional tactile sensing. The SMF is composed of two parts: a ‘finger’ covered with a skin-like flexible sheath containing five liquid metal (LM) coils and a ‘phalangeal bone’ containing a magnet. Due to the changes in magnetic flux through the LM coils caused by external forces, diverse induced voltages are generated and collected in real-time, which can be explained by Maxwell’s numerical simulation. By the analysis of the signals generated by the five LM coils, the SMF can detect forces in varied directions and distinguish 6 different common objects with varied Young’s moduli with an accuracy of 97.46%. These capabilities make the SMF suitable for complex unknown space exploration tasks, as proved by the black box exploration. The SMF can enable the development of self-generated-signal and multidirectional tactile perception for future robots.

具有多向触觉感知的机器人手指对于机器人探索复杂的未知空间非常重要，特别是在无法可视化的环境中。不幸的是，大多数现有的机器人手指触觉传感器无法检测多向力，这极大地限制了它们在复杂环境中进一步发展的潜力。在这里，我们展示了一种软磁电手指（SMF），它可以实现自生成信号和多向触觉传感。SMF 由两部分组成：覆盖有皮肤状柔性护套的“手指”，其中包含五个液态金属 (LM) 线圈；以及包含磁铁的“指骨”。由于外力引起流经LM线圈的磁通量发生变化，实时产生并收集不同的感应电压，这可以通过麦克斯韦的数值模拟来解释。通过分析五个LM线圈产生的信号，SMF可以检测不同方向的力，并区分6种具有不同杨氏模量的不同常见物体，准确率高达97.46%。这些能力使得SMF适合复杂的未知空间探索任务，正如黑匣子探索所证明的那样。SMF 可以为未来机器人开发自生信号和多向触觉感知。