Current Sensing Teaches a Robot to Accept Touch: Compliant Control in Practice with Mino

When Robots Need Gentle Handling

A developer shared on Reddit how they added current sensing to their homemade robot, Mino. The problem was concrete: they wanted to pet their robot like a dog, but the servo would push back — which felt unnatural and put unnecessary stress on the hardware.

The solution was straightforward but effective — integrating current sensors directly on the PCB and writing an algorithm on the MCU to detect external forces on the servo in real time. When the detected force exceeds a threshold, the servo enters "follow mode" and moves with the external force instead of resisting it. You can see this behavior in action around the 0:12 mark in the video.

How Current Sensing Enables Compliant Control

Reading External Force from Current

When a servo resists an external force, it draws more current — a direct physical relationship. By placing a current sensor in the power supply circuit, the MCU can approximate the load the servo is currently experiencing. An abnormal spike in current readings signals that an external force is pushing against the joint — like a hand petting the robot.

The elegance of this approach is that it requires no dedicated torque sensors or expensive encoders. It uses current — a signal already present in the system — to indirectly sense contact force. For low-cost DIY robot projects, this is an excellent bang-for-your-buck solution.

In practice, current sensing is typically implemented using Hall-effect sensors (such as the ACS712 series) or low-resistance shunt resistors paired with op-amps. The former requires no circuit interruption and offers good isolation, making it suitable for monitoring multiple servos from a single sensing point; the latter is cheaper but requires attention to common-ground issues. The MCU reads an analog voltage converted by an ADC, so the software layer must also handle noise filtering (e.g., moving average or low-pass filtering) and baseline calibration — a servo's no-load current drifts with posture, temperature, and load, meaning robust threshold tuning often demands more debugging experience than hardware selection.

The Dual Value of Follow Mode

When the external force exceeds the set threshold, the servo switches to follow mode. This delivers two distinct benefits:

First, it improves the interaction experience. Instead of rigidly resisting, the robot "cooperates" with the user's movement, making touch-based interactions — like petting — feel natural, and reinforcing the robot's emotional identity as a "pet."

Second, it protects the hardware. A servo that continuously fights external force will overheat and wear down mechanically. Follow mode is, at its core, an overload protection mechanism. The developer put it plainly in the original post: "Best spent extra lines in the BOM and the code."

Why This Small Feature Deserves Attention

Compliance control is not a new concept in industrial or service robotics, but it typically relies on complex force-feedback hardware or high-precision servo systems. This project demonstrates how to approximate the same effect on a hobbyist robot at minimal cost.

For consumer-facing robots increasingly designed around human-robot emotional connection, "how the robot responds to touch" is a surprisingly underrated design dimension. A robot that pushes back instinctively keeps people at a distance; one that yields to contact invites it. The layer of compliance that current sensing provides is precisely the kind of detail that closes the gap between humans and machines.

From an engineering perspective, this project also exemplifies sensor reuse — solving a problem that seemingly requires dedicated hardware by leveraging an already-available signal (supply current). That kind of resourcefulness is exactly what makes the DIY robotics community so creatively valuable.

In academic terms, compliance control is divided into active compliance and passive compliance. Passive compliance uses elastic elements to absorb external forces at the mechanical level — like the spring-buffer structures used in early Boston Dynamics robot legs. Active compliance uses sensors to detect force or torque in real time, with a controller dynamically adjusting motor output to "give way." Industrial-grade active compliance typically requires six-axis force/torque sensors (costing thousands to tens of thousands of dollars) or high-resolution encoders paired with impedance control algorithms. Mino's current sensing approach is essentially a simplified implementation of active compliance: it uses current as a proxy signal for force, sacrificing precise force quantification but preserving the core logic of "detect abnormal load → change control strategy." Within cost constraints, it achieves a result that is "good enough." This kind of dimensional reduction and substitution is extremely common in embedded and DIY contexts, and understanding it is key to appreciating the value of this approach.

Takeaway

Mino's upgrade is small in scope but a great example of what details can do: what makes a robot feel truly "alive" is often not raw compute power or degrees of freedom, but how it responds to being touched. Current sensing combined with follow mode trades a minimal investment for better interaction quality and longer hardware life — a genuinely worthwhile tradeoff.