The whole system consists of the Pioneer 3 DX (see Fig. 5.3) containing an onboard PC (see [Ver10]) and an additional laptop where the ILCA architecture is installed. This structure is simply motivated by some advantages regarding the evaluation of exper- imental data and a clear separation of technical system and cognitive architecture as well. On the onboard PC, a control program with several filter and controller routines runs, which communicate (partially) over a micro controller with the robot’s sensors and actuators. The technical communication between control program and cognitive architecture is realized by a WiFi connection. Therefore, the robot is equipped with a network card and an antenna, which restricts the robot’s operation area to a certain distance to the laptop’s location. However, this operation area is large enough for the chosen example scenario. Furthermore, the operation area can be enhanced arbitrarily if the cognitive architecture it is running on the onboard PC.
© SRS 2010 Sonar Camera RFID Lightbarrier Touch sensor Laser Motor/wheel Gripper Speaker WiFi
Figure 5.3.: Sensors and actuators of the mobile robot
The mobile robot contains three lead-acid secondary batteries, each with a capacity of 7,200 mAh. Depending on the scenario to be realized and the batteries state-of-health, the robot can perform approximately two hours independently from an external energy supply. However, a complete charging takes much longer, especially if the onboard PC is running. Hence, the robot has to be charged in regular periods to realize longer oper- ation times (which automatically reduces the batteries’ lifetime extremely).
The system is mechanically designed for the navigation in office environments and it is also equipped with several sensors and actuators to perform related tasks. In order to move transversally and radially in a building, the robot contains two large wheels mounted laterally and a small freemoving wheel at the back side. The large wheels can be controlled independently from each other by two motors, whose rotation is measured by odometry sensors, which are applied to estimate the driven way roughly. Due to several influences as slipping wheels or measurement errors, the internal representation of the system’s position in the environment has to be adjusted continuously. Possible solutions for this are probabilistic approaches estimating the robots position from a geo- graphical map and distance measurements. In order to measure the distance to other objects, the robot is equipped with 16 ultra sonic sensors and a laser range finder. The ultra sonic sensors are mounted around the robot and they can be used to measure the rough distance to obstacles on different heights. Hence, also chairs and tables can be detected. In contrast to the ultra sonic sensors, the laser range finder can detect the distance to objects very precisely, but only on a certain height and only at the robot’s
Chapter 5: Realization of Cognitive Technical Systems 102
front side within a range of 180◦
. Besides the distance, also the color and the shape of objects can be measured by a controllable video camera (see [Mob04]). Finally, objects can be gripped and lifted by a gripper (see [Mob07]) mounted at the front of the robot, which contains two light barriers and a touch sensor. Hence, the robot can carry objects if they have a suitable size and weight.
Besides the sensors and actuators provided with the Pioneer 3 DX system originally, the robot is equipped with additional components as a RFID-reader, an I/O-card and two speakers. The RFID-reader generates an electromagnetic field and detects whether a transponder-tag is within this field. The used transponder-tags are smart labels which can be read or written. Hence, the RFID-reader can be used to detect fixed landmarks in order to estimate the location of the robot. Alternatively, also a probabilistic approach (included in ARNL [Mob09], an API provided by the robot’s manufacturer) could be applied. However, due to the fact that this approach makes use of static geographi- cal maps which do not represent a dynamical changing environment, the localization through landmarks is chosen. The RFID-reader is connected with the onboard PC by a separate USB-port. Depending on the orientation of the RFID-reader, transponder-tags within a range of up to one meter can be detected. However, since the RFID-reader is mounted on a metallic body containing several devices with electromagnetic radiation (disturbing the electromagnetic field of the RFID-reader) and the robot is moving dur- ing measuring, the measuring range is reduced extremely. Hence, the RFID-reader is located beneath the robot in order to detect transponder-tags on the corridor’s ground with a high probability.
To realize simple interaction with humans, an additional I/O-card and two speakers are added to the system. The speakers can be used to generate acoustic signals, e.g., to warn passersby. Furthermore, simple statements or questions can be addressed to humans. In order to answer the robot’s questions or to give certain information (e.g., to define a goal), two buttons are mounted on the robots back side and connected to the I/O-card. The I/O-card is connected again with the onboard PC’s USB-port and contains eight digital inputs, eight digital outputs, and eight analog inputs. Hence, the connection of further sensors and/or actuators with the robot is relatively simple.