|Size:||2,96 m x 1,84 m x 0,32 m (locomotion posture)|
|Weight:||Approx. 236 lbs|
|Power supply:||LiPo primary battery: 45,6 V and 10 Ah|
|Speed:||Approx. 2,2 mph|
- Legs: 4 x 6-DOF
- Arms: 2 x 6-DOF
- Hands: 2 x 8-DOF
- Torso: 5-DOF
- Head: 4-DOF
- Rotational actuators within the arms, legs and head: Brushless DC motors with Harmonic Drive gear
- Linear actuators within the legs: Brushless DC motors with
- Linear actuators within the torso: DC motors with planetary gear and trapezoidal screw drive
- Laser range finder: Hokuyo UTM-30LX
- Stereo camara system: 2 x Procillica GV 2450C
- IMU: -IMU: 2 x XSens MTi
- 122 temperature sensors
- 191 current sensors
- 12 tactile sensor arrays with 40 sensing elements
- 88 rotational absolut encoder
- 14 six axes force-torque-sensor
Independent LVDS chains for the control of both arms, the four legs, the torso and the head.
DFKI electronic stack for all drives:
- Input voltage: 12V-54V
- FPGA- Spartan 6:XC6SLX45
- Serial communication for Spartan-6 (320MSym/s)
- Sinusoidal commutation
- 2 x LVDS for local sensors
- 2 x Ports for IC Haus MU sensors
- 4 x Status LED via plug
|Grant number:||This project is funded by the German Space Agency (DLR Agentur) with federal funds of the Federal Ministry of Economics and Technology in accordance with the parliamentary resolution of the German Parliament, grant no. 50 RA 1218.|
|Team:||Team I - System Design|
SAR- & Security Robotics
Logistics, Production and Consumer
Behaviors for Mobile Manipulation (05.2012- 07.2016)D-Rock
Models, methods and tools for the model based software development of robots (06.2015- 05.2018)LIMES
Learning Intelligent Motions for Kinematically Complex Robots for Exploration in Space (05.2012- 04.2016)
Expandable Rover for Planetary ApplicationsCharlie
Machina Arte Robotum SimulansPhobos
An add-on for Blender allowing editing and exporting of robots for the MARS simulationRock
Robot Construction Kit
MANTIS is a multi-legged robot with six extremities. The system was developed as a platform for interdisciplinary research in the area of mobile manipulation with multi-legged robots. To fulfill a variety of different tasks the robot is capable of operating in two different postures. In the manipulation posture MANTIS uses the four rear legs for locomotion and the two front legs for manipulation. Furthermore, in the locomotion posture the robot walks on all six extremities, which is a big advantage in difficult terrain. This flexibility allows to solve complex scenarios with only one system. Several software components as well as planning methods can be evaluated within such scenarios.
MANTIS is designed as an autonomous system with an integrated power source, a Mini ITX computer and the DFKI ZynqBrain which serves as central control unit for the robot as well as several FPGA-processing units and microcontrollers distributed across the robot are used for decentralized data pre-processing and control of subsystems. For the purpose of perception, the robot is equipped with pressure sensors in the sole of each foot, tactile sensor arrays in the fingers, a stereo camera system, a scanning laser range finder and inertial measurement units in the head and abdomen. The system is furthermore equipped with current, voltage, temperature, acceleration, position, force and torque sensors to monitor the internal status.
The robot MANTIS is being developed within the project LIMES. This research aims to generate and optimize different locomotion behaviors for different situations. Machine learning methods are used to learn different behaviors which are tested in the simulation environment MARS and evaluated afterwards on the real system. The acquired set of locomotion behaviors is used and adjusted one the real system to adapt to the current environment and tasks.
In the project BesMan, the system is used to study two-arm manipulation behaviors, which were generated by the use of new methods for imitation learning.
In the D-Rock project, MANTIS will be used as a reference system to demonstrate the capabilities of a model-based development approach developed in the project. To demonstrate the effectiveness of this model-based approach the robot will be used to solve a DARPA Robotic Challenge scenario.