Robotic Control Same-Cycle Motor Control for 18-Motor Humanoid Robotic Hand

Anbieter zum Thema

Inova Semiconductors GmbH

FAULHABER Antriebssysteme

ROHM Semiconductor GmbH

At electronica 2026, Inova and Lattice demonstrated a humanoid robotic hand control platform that synchronizes 18 motors and multiple sensors within a single 100-microsecond control cycle.. APXpress networking and Lattice Nexus-based FPGAs enable synchronized 10 kHz motion control with communication latency below 10 microseconds, supporting precise motor coordination and simplified sensor connectivity.

Developed in collaboration with the German Aerospace Center (DLR) Institute of Robotics and Mechatronics, the platform combines Inova Semiconductors’ APXpress high-speed networking technology with low-power Lattice CertusPro-NX and CrossLink-NX FPGAs. The architecture integrates 18 motor controllers, joint-angle sensors and a tactile array comprising 10 sensors into a single deterministic communication network. DLR contributes its expertise in robotic manipulation and field-oriented motor control, while the FPGA-based implementation provides distributed processing capabilities at individual motor nodes.

A key technical achievement is the completion of the distributed motion-control loop within one 10 kHz control period, equivalent to 100 microseconds. Communication between sensor sampling and the delivery of updated target values takes less than 10 microseconds, allowing all motor axes to respond synchronously. Conventional cyclic architectures typically introduce a delay of one or two control periods between feedback acquisition and current adjustment. By eliminating this inter-cycle delay, the demonstrated system supports faster control responses and more stable impedance regulation, particularly for robotic applications involving physical human interaction.

Field-oriented control (FOC) is implemented locally on a Lattice FPGA at each motor, operating at a frequency higher than the network's 10 kHz control rate. This distributed processing architecture enables each motor to regulate its electrical currents independently while receiving synchronized target updates through the APXpress network. Coordinating the motor axes within the same control period reduces timing differences between fingers, supporting smoother movement and more precise manipulation. The combination of local high-speed current regulation and deterministic network communication is particularly relevant for multi-axis robotic hands requiring tightly coordinated motion.

The network also accommodates joint-angle measurements and tactile feedback through a separate communication schedule designed to protect time-critical motor-control traffic. Although tactile sensing generates a substantial portion of the transmitted data, scheduled transfers prevent interference with motor commands. This separation allows the system to consolidate motion control and sensor communication onto the same physical network without compromising deterministic timing. The available APXpress bandwidth also eliminates the need for an independent sensor wiring harness, reducing system complexity while maintaining access to multiple feedback sources.

To minimize cabling, the motor controllers are arranged in a daisy-chain topology using short point-to-point connections instead of a conventional star configuration. This approach reduces the number and length of wires routed through the robotic hand, helping limit weight and mechanical restrictions around moving finger joints. The compact network arrangement also improves integration flexibility by reducing bulky cable bundles in mechanically constrained areas. Combining the motor and sensor connections into one network provides a more scalable approach to distributed electronics in articulated robotic mechanisms.

Beyond the demonstrated hand configuration, the APXpress network topology has been designed to accommodate additional data streams from cameras, radar and other perception sensors. Inova is developing silicon solutions around this modular architecture, with Lattice supporting FPGA integration and DLR providing scientific expertise in robotic control. Although the presented system remains a development demonstrator, its same-cycle control performance, distributed FOC implementation and reduced wiring requirements establish a technical foundation for more responsive, compact and highly integrated humanoid robotic systems.(sg)

(ID:51014452)