uBot-4 Hardware Subsystems:

  • Motor control
  • Inertial measurement (IMU)
  • Balancing controller
  • Human Robot Interface (HRI)
  • Sensing/Perception
  • Technical Specs.

Motor Control

Currently, the uBot-4 uses 1 DSP to control the 2 base wheels. The DSP receives Kalman filtered data from an Inertial Measurement Unit (IMU) and implements a digital closed-loop controller for calculating the required torque at each base wheel in order to stabilize the platform (LQR,state space model, inverted pendulum). There are 9 Gamatronix Gamoto PID motor controllers to drive the shoulder rotation and arms. The 9 Gamotos are connected, via two shared serial ports (115.2kbps), to a StrongARM based ADS Bitsy SBC running Linux. To control the uBot-4 a host PC connects over 802.11 through a custom GUI. The GUI presents the user with robot's configuration and is able to update the robot's motion and state information via teleoperator interface (joystick).

IMU

The IMU serves the purpose of implementing a Kalman Filter to transform samples of two analog signals (MEMS accelerometer and rate gyro) into accurate values of tilt angle and angular velocity about an axis tied to the base wheel's axle and parallel to the floor.

Balancing Controller

The DSP updates the base wheel torque approximately XX times per second (the IMU may only generate XX new updates per second). The torque is calculated by the formula: 

torque = K1*position + K2*velocity + K3*tilt + K4*rate_of_tilt

The values for K1, K2, K3, K4 are determined by solving for the optimal gains for the system state space (linear differential equations) and a quadratic cost function. This type of feedback controller is called a Linear Quadratic Regulator (LQR).

The balancing controller introduces a passive impedance such that the system tends to move away from external forces.

HRI

The uBot-4's small form factor and impedance characteristics enable direct and physical interaction with humans and within human centric environments.

Sensing and Perception

  • Smart room
    • Navigation and localization
    • Object localization for grasping
    • Visual servoing within bimanual workspace
  • Contact forces via force sensing actuators

Technical Specifications

The uBot-4 was designed in-house using CAD and assembled by gleeful graduate students. Machining was contracted out through mfgquote.com.

  • 11.5kg
  • Power supply 14.4V 7.6Ahrs
  • Average run-time 1-1.5hours
  • Top speed approximately 3mph
  • Payload at end-effector 5kg/arm full extension
  • 0.5m arm length