Competition Year: 2017
University: Zhejiang University
Team Members (with year of graduation): LI Chao, MO Xiaobo, LI Chuan(2017), ZHOU Kun(2017)
Faculty Advisers: ZHU Qiuguo
Email Address: email@example.com
Quadruped robot is one of the main research highlights of legged robot, which has significant advantages in complex terrain situation. Unlike current research project, namely boston dynamics bigdog which uses hydraulic system, Chitu is totally powered by electrical DC motor to achieve high speed control with least noise and fast system response. (see Figure 1). It can walk, trot and run through steps, slopes and other unknown complex pavement brick.
cRIO and LabVIEW were applied in this project. Thanks to the flexibility of graphical programming language LabVIEW, the whole project was finished in 6 months. Detailed specification could be found in Form 1.
Figure 1, Chitu, Quadruped robot.
Form 1, Specification of Chitu Quadruped robot
Software: LabVIEW 2014 SP1, LabVIEW Real-Time Module, LabVIEW FPGA Module, IndCom_ECAT, NI-VISA, NI RIO
Hardware: NI cRIO-9033, NI 9201
Other: 12x BLDC Motor, 12x High-Precision Encoders, 12x Driver, 4x Force Sensor, 1x Xsens IMU
In several situations of complex road environment, traditional wheeled robot is not able to meet the high requirements of flexibility, especially in forest or ruins full of sticks, steps, slopes and brick, which always happens in rescue after disaster.
This project aims to solve the problem of complex environment transportation, especially for civil engineering or other related complex environment exploration tasks. For example, the transportation of precision instruments and devices in ruins are usually done by humankind, this project will largely benefit the related tasks and also contribute to the unmanned exploration. In addition, the traditional gas-powered robot cannot be applied in an environmental friendly forest, which might lead to exhaust pollution. While the totally electronics-powered Chitu could fits this situation with the least air or noise pollution.
Transportation in complex environment. Pic from NEWSSC.ORG
To solve these problems and reduce the environmental impact. Chitu was designed.
Motion control of Quadruped Robot within unknown complex terrain
This project presents a motion control method for quadruped robot within unknown complex terrain. The method is based on stability criterion of zero moment point (Zero Moment, Point, ZMP), aiming to achieve a real-time trajectory planning. Based on the swing leg landing strategy and algorithm deployed for perception of unknown terrain information, the real-time controller could achieve an adjustment of the support leg length to control the robot posture to match the terrain parameters. This project has already achieved an adaptive control of terrain height and slope.
Fig. 2 Quadruped Robot crawling model and support domain
A dew quadruped robot model and control method based on vertical virtual spring was designed and implemented. To achieve a high precision control and correct the rhythm shift in the robot motion, the phase calculation and phase feedback are introduced.
Figure 3 simplified model and double vertical spring model
cRIO as the client server and embedded controller
This project was powered by CompactRIO-9033. The communication and electrical interface was shown in Figure 4 and Figure 5. CompactRIO-9033 communicates with the 12 joint drivers through the EtherCAT interface, obtaining motion information of each joint, and sending controls commands separately; Pose and motion data of IMU was sent to cRIO through RS485 interface; By measuring the pressure of Chitu`s foot using NI 9201, the whole sensing picture was completed. All the devices are electrical-powered with no hydraulic sub-systems, this will maintain a faster response and more environmental friendly experience.
Figure 4, System structure including host and embedded controller
Figure 5, CRIO interface
PC as the host and data recording device
By communicating with the host system through wireless connection, this project achieves a more stable and faster data recording and real-time processing process. RTOS running in cRIO also provides a precise timing reference for the algorithm design and DAQ. Figure 6 is a PC monitoring interface of Chitu.
Figure 6, Host PC monitoring system
Fiture 6 Chitu `s status in different road condition
Why Chitu is special
The quadruped robot has a strong adaptability to the environment, and it can flexibly deal with the complicated road environment which cannot be adapted to all kinds of wheeled robots. Powered by LabVIEW and cRIO, control algorithm could be deployed to enhance the operation ability.
Chitu has shown strong ability in passing complex road environment and fast running. But what makes it special is listed as follows.
Chitu has 12 joints to power this complicated and precise system. For each joint, the controller should collect the information of angle, angular velocity to fits system model, based on which the coordinated commend was generated by the algorithm deployed in the controller.
How to deal with the large amount of data is the key in this project. Host PC need to pick up the right routes, and the target need to complete the signal processing and conditioning, including stochastic estimation, high order metrics calculation and sensor fusion algorithm implementation to better estimate the robot status. In addition, the high speed and real-time requirements of the control system requires the controller to complete the control cycle in a very short time. Considering the large data processing and fast control response requirements, we chose the CompactRIO as the controller for Chitu. CompactRIO provides IndCom_ECAT, which allows us to quickly establish a EtherCAT based CompactRIO to achieve multiple drives` high-speed real-time communication. The flexibility of cRIO also leaves us an infinity space to update Chitu to fits multiple environmental conditions.
In order to pass the unknown complex pavement, Chitu needs to get the real-time body attitude and motion information through the embedded IMU. The contact condition was obtained by the pressure sensor located under the feet of Chitu, and through the GPS system, the outdoor location information was transferred through the establishment of the host computer and the control system.
To make the whole system more flexible, Chitu was designed based on cRIO, which provides a rich I/O and multiple modules supporting different communication interfaces, including: EtherCAT, RS485, RS232, USB and wifi. Modular design is also applied in this project. Each module could be updated based on the application and sensors or power modules. One could even design different suites to fits the changing road situation, the only thing you need to do is to replace the suites and config the right control algorithm for different road situation.
In the early developing stage, the developer might inevitably encounter some problems need investigation and further debugging, thus improve the performance. This is essentially important in controller design process. For different algorithm development, the tuning method is so important. For example, in PID controller design process, the parameters of P, I and D need to be test out by tuning to fits the system model. While in most cases, this tuning process should be done offline because of the limits of hardware. While cRIO could overcome this shortage by providing the customer with a real-time tuning platform. In LabVIEW, it will be quite straight forward for the developers to debug and find the most proper controller setting in real-time mode, which could save product development time.
Chitu was designed for scientists and researchers to do legged robot research. High flexibility of the platform and the rich interface leave enough space for the scientist to do more on it. Also this totally electrical solution could help the further modularity development of the suites ecosystem, which aims to accelerate the innovation of control algorithm design, smart sensing and robotics related research.