2019/02/19

Cobot

Dual-encoder actuator case

As human-robot collaboration becomes increasingly prevalent, our expectations for robots have also risen to a new level. Traditional large, bulky industrial robots—such as those used for palletizing, handling, and welding—can only operate within safety cages and are not suitable for working directly at our workstations.

Collaborative robots, known for their light weight, safety, and ease of use, are ideal for working with humans, especially in light industry. They handle delicate tasks more flexibly, and as collaboration standards rise, so do the requirements for their precision, size, and safety.

Actuator

For robots, every joint movement relies on:the motor, gearbox, encoder, and driver.Traditional large robots have relatively low space requirements. To save costs, they are typically assembled by directly connecting many of the aforementioned modules with different functions.

Collaborative robots have higher space requirements, driving the need for compact, integrated modular actuators at each joint. This enables efficient layout, easier production and maintenance, and better cost performance.

As shown above, each joint uses an integrated actuator to carry out commands from the central controller. Beyond torque, collaborative robots require higher standards in the following areas:

• High precision:The motion accuracy of a collaborative robot depends on the precision of each actuator. Even with high-precision mechanical components, achieving overall accuracy requires precise position feedback from the encoder within each actuator.
• Safety:Collaborative robots must be able to sense external forces—just like feeling your way in the dark, where the arm stops upon detecting an obstacle. This forms the basis for collision safety detection in robots.
• Ease of use: Collaborative robots must be easy to operate without complex procedures or specialized skills. Like being led by the arm while blindfolded, robots use force-guided teaching to simplify use and improve collaboration.

For safety and ease of use, robots need to sense external forces on their bodies. At each joint, this means sensing external torque—better torque sensitivity allows detection of smaller torques, providing optimal performance feedback.

This article focuses on the described actuator solution. The image above shows a case example of the solution, with detailed information provided below.

Torque sensing solution

Existing torque sensing solutions include:
• Current: Estimates external force via motor coil current; easy but low accuracy.
• Dual encoder: Adds a high-resolution encoder to measure output position changes for torque calculation; low cost and high accuracy.
• Torque sensor: Measures torque at the output; high cost and complex algorithms.
• Strain gauges: Multiple strain gauges attached to the harmonic flexspline to measure torque; highly complex structure and difficult calibration.

Dual encoder issues

Speaking of dual encoders, since an additional encoder is installed at the output side as shown above, what challenging issues arise from this design that make the solution impractical or difficult to implement? The main points are as follows:

• Position of the encoder relative to the output flange:To measure rotor movement, the encoder’s stator must be fixed to the actuator stator, requiring installation near the output flange on the stator side. This complicates the structure, lengthens the output shaft, increases bending moments, and reduces structural stability.
• Encoder wiring:Since the encoder is on the actuator rotor side, power and communication lines must pass through a slip ring to the other side, increasing the slip ring’s channels and size, resulting in design complexity.

Dual encoder

To solve these issues, the preferred solution uses a hollow shaft design (shown above), moving the output-side encoder to the actuator stator side. A return shaft passes through the hollow shaft to provide rotation data, addressing encoder mounting and wiring challenges.

The above image shows the structure of this solution. Encoder 1 drives the motor, while Encoder 2 measures the output position changes.

The image features KingKong Technology PCB series magnetic encoders, providing sufficient resolution and accuracy. Their ultra-thin design enables a highly compact structure, with both encoders occupying only about 15mm axially.

Dual encoder algorithm

该方案中一般使用的谐波减速比为几十至100左右的大速比

• At static state:When external torque acts on the output, encoder 2 detects slight changes while encoder 1 remains unchanged due to high reduction. Comparing them reveals subtle user forces. The driver then either lets the motor follow the force (if teaching mode is on) or compensates to keep the position fixed.
• In dynamic state:After obtaining data from motor-side encoder 1, the theoretical output shaft position is calculated and compared with output-side encoder 2 data to determine external force, which is then passed to the driver for control.

Unlike current or torque sensors that measure changes over time, dual encoders measure torque based on spatial position. This provides more accurate external force detection during motion, as time-based signals fluctuate due to changing arm dynamics.

In dual encoder setups, spatial measurements from the encoders are typically combined with current-based time measurements to achieve better output data.

In the future

KingKong Technology is committed to providing better solutions. Our patented product, shown in the topology diagram above, offers a more compact design (about 10mm in height), improved wiring, and control solutions. It will be released soon. Please contact us for the latest updates.

This is a case study of a dual-encoder actuator in collaborative robots. If you have any questions or need assistance, please feel free to contact us.