SLIM - A Scalable and Lightweight Indoor-navigation MAV as research- and education platform

Fig. 1: The SLIM hovering in front of a victim during a search and rescue competition of the camera drones lecture.


Indoor navigation with micro aerial vehicles (MAVs) is of growing importance nowadays. State of the art flight management controllers provide extensive interfaces for control and navigation, but most commonly aim for performing in outdoor navigation scenarios. Indoor navigation with MAVs is challenging, because of spatial constraints and lack of drift-free positioning systems like GPS. Instead, vision and/or inertial-based methods are used to localize the MAV against the environment. For educational purposes and moreover to test and develop such algorithms, since 2015 the so called droneSpace was established at the Institute of Computer Graphics and Vision at Graz University of Technology. It consists of a flight arena which is equipped with a highly accurate motion tracking system and further holds an extensive robotics framework for semi-autonomous MAV navigation. A core component of the droneSpace is a Scalable and Lightweight Indoor-navigation MAV design, which we in order call the SLIM. It allows flexible vision-sensor setups and moreover provides interfaces to inject accurate pose measurements form external tracking sources to achieve stable indoor hover-flights. The purpose of this website is to present details about:

  • Parameters for hover-flight time estimation
  • Components of the high-level ROS control framework
  • Assembly details and instructions

It is important to mention, that this website currently provides supplementary information only, whereas it is planned that a complete and detailed description of the SLIM platform will be made availabe until June 2019.

Parameters for Estimating Hover-Flight Times

Amongst others, important boundary conditions for the SLIM design included a maximum all-up-weight (mAUW) of 700g, a Battery with 3 cells (3S, 11.1V) and 2300mAh of capacity and a minimum required flight time of 10min. In order the the hover-flight time was estimated. In general, the hover-flight times of a Lithium-Polymer (LiPo) battery powered quadcopter, can be calculated based on the following parameters:

  • C defined as the battery capacity given in [mAh].
  • Vn defined as the nominal battery voltage given in [V]. Considering LiPo batteries, the nominal voltage is defined with the cell voltage (Vc=3.7V) times amount of cells S, in order Vn=S*Vc.
  • Pm defined as the electrical power given in [W], required by one engine to lift a quarter of the total weight (mAUW=0.7kg) of the MAV, since the SLIM setup includes 4 engines in X-configuration. It is further assumed that all 4 engines and props are identical and the center of gravity is roughly situated in the middle of the used frame. Beforehand an estimate of Pm,est. was calculated, whereas as a proof of concept a true value Pm was empirically measured during hover flight, assuming the worst case of mAUW=0.7kg.
  • Pe is defined as the electrical power given in [W], which is required by the electronic components of the SLIM As a worst case assumption, Pe was estimated as the maximum power consumed by the SBC (PSBC=5V*4A=20W) and the Flight Controller (PSBC=5V*1.0A=5W). An addition of 1.0A at 5V supply level was considered as safety margin (Pmargin=5V*1A=5W).
  • η can be defined as an efficiency-factor that takes into account energy losses from various sources. One major source is to keep the LiPo battery voltage level above a certain threshold when discharging. Another important source are losses from wires and electronic components. As a rule of thumb η is most commonly estimated with 0.8 for typical MAV configurations.

Framework Components for High-Level Control (ROS)

In this section, the components of the high-level control framework are represented. The high-level framework is implemented in ROS whereas an overview of the components is given in Fig. 2.


Fig. 2: Core components of the SLIM software-framework.


This section explains more details about parts list and assembly instructions of the SLIM.

Assembly Instructions

Fig. 3: Principle layout for the placement of main components of the SLIM. The custom mounts are shown in blue, whereas electronic components are represented in green colors. The legs attached to the base frame are optional for the RTF setup.

In the following detailed instructions for assembly of an RTF configuration are discussed. While Fig. 3 represents the stack-like structure of the physical setup of the SLIM, the according parts list with the part ID is given in Tbl. 1.

Engines (ID6), Propellers (ID5) and connector cables are mounted with the according fasteners in stock configuration to the frame.

Starting from the bottom, the SLIM is designed to land and takeoff on the attached battery. This has two advantages as it saves additional weight of the frames legs and also makes it easily attach-/detach-able for recharging. The battery (ID10) is fixed to the custom battery-bay mount (ID2), part of the Bebop 2 frame (ID1), via a Velcro-Fastener (ID18).

As the Bebop 2 frame provides a hollow space in the center, it makes placement of smaller sized components, like the SBEC (ID13) and the RC-Receiver (ID9), more efficient.

On top of the stock Bebop 2 frame, we attach another custom extension mount (ID3). It is responsible for holding the ESC (ID7), the Flight Controller (ID12) and the SBC (ID13). The ESC is mounted with screws (ID17) at the backside of the frame, whereas the Flight Controller is mounted in the same way and centered in the middle. The SBC is mounted above, via 4 hexagonal spacers (ID18). Finally, the SBC is attached with a serial connection to the flight controller (ID16), a WiFi link (ID14) and one additional vision sensor via USB.

The third custom mount (ID4) is optional and able to hold the additional vision-sensor, for example an RGBD-Sensor (ID15).

Parts List

This section discusses a full RTF setup of the SLIM, including the Hardkernel Odroid XU3 single-board-computer and the ORBBEC Astra sensor configuration to achieve online environmental mapping during flight. The full parts list, with according weights, required quantity and availability for ordering, is shown in Tbl. 1. The presented setup can be estimated with an overall weight of ~613g which is below the required weight, including safety buffers for potentially more payload.

Tbl. 1: SLIM Parts List For Online-Mapping Configuration including an RGBD-Sensor.