Statement of problems the project address

In robotics, the detection of physical 3D environment has been especially treated using very sophisticated sensors such as scanning laser range finders or cameras coupled to the infrared. These types of detection are very expensive for the former and heavy in terms of treatment for the latter. They are not adaptable to all robotic systems. Indeed, the dense stream of data provided by a camera includes unnecessary mapping information. The result provided a 3D environment very accurate, giving us many details. We believe these solutions give us too much information, and they can not be processed by on-board systems, such as robots, which have another function mapping. For example, a rolling platform for domestic surveillance might work by 3D mapping: This solution would have an overall vision of the place to watch without having any blind spot. Our targets are robots that might need to know about their 3D environment but which do not benefit from computing resources and sufficient memory to accept these mapping solutions.

Tools

ROS

Robot Operating System (ROS) is a collection of software frameworks (Robotics middleware) for robots. ROS provides standard operating system services such as hardware abstraction, control of low-level devices, setting work of commonly used features, the transmission of messages between the processes, and the package management.

WIFI

The Wi-Fi is already in place with any GNU / Linux ARM. Moreover, ROS runs natively on the Ethernet protocol. We bought a router in order to have our own WiFi.

RaspberryPi

This card is very full, it allows us to do many things, as well as the upper low. In addition, network management is already implemented by the OS. We have to go on a RPi 2 Model B because its characteristics are those that suit us the most: CPU: 900 MHz quad ARM Cortex-A7 (ARM v7) Memory (SDRAM): 1 GB Number of USB 2.0 ports: 4 Network Connectivity : RJ45 port Maximum consumption measured: 350 mA Weight: 45 g

Hardware

Regarding the engine, we have set some constraints. The first was explained in the section on objectives. It concerns the pitch angle. Indeed, we want a satisfactory accuracy in our mapping, we must have a small angle measurement to stick to our goals. However, it is also necessary that these measurements are made in less than 3 minutes to not interfere with the user duration. The second constraint relates to the weight supported by the engine. The whole structure will rest on, so we choose the material according to its holding torque: it had to be high enough to support all the other components. We therefore chose an engine that meets these requirements: Engine 23HS41-1804S

Stepper motor-no bipolar Nema size 23,

High precision, vibration and very low noise levels. No angle: 1.8 ° Power: 4.95 V Resistance / phase current 2.75 ohm / phase Inductance 1.8 A / phase Holding torque mH 17: 24 kg.cm connection: 4 Dimensions of son axis: Ø 8 x 20 mm Size: 57 x 57 x 105 mm Weight: 1.25 kg

For the actuator, we needed one that rotates 180 degrees: RB-Hit-87

HS-485HB offers optimum torque as well as a resolution and a perfect centering Speed ​​(sec / 60o): 0.20 (4.8 V), 0.17 (6 V) Torque (kg-cm): 5.2 (4.8 V), 6.4 (6 V) Dimensions (mm): 39.9 x 19.8 x 37.9 Weight (g): 45

The step-down controller is necessary for 5V supply components.

Its intensity must exceed the total need of the components. We have chosen a voltage regulator Step-Down: D15V70F5S3 Input voltage between 4.5V and 24V Output voltages selectable 5V or 3.3V 7A When we use the engine with a PWM, we generate a spurious frequency. In addition, the engine running itself generates noise. For these two reasons, it is often useful to add filter capacitors across the motor terminals.

LIDAR

We agreed on 2 degrees between two measurements. Indeed, it seems to be a good ratio between the scanning time (216S to 120x360) and the gap between the points according to their LIDAR-object distance. If we can get a small gap we will, this would increase the accuracy of measurements, however, we must also respect the time constraints of up to 3 minutes for all the measures.

3D printing

3D objects may be designed by modeling software number through a long process. This prototyping process is accelerated by turning to Blender. Indeed, modeling our structure will be made on FreeCAD but to verify that it matches what we expect, we will check it with Blender. To address the problem of heat dissipation components, we decided to create our way of honeycomb structure to allow heat to dissipate.