A Brief and Concise Introduction to Aerial Manipulators

by Julio Mendoza

Definition and characteristics

An aerial manipulator is a kind of robotic arm capable of flying. It is basically composed of two devices, or a fusion of the two: one or more robot manipulator, and one or more aircraft.


There are currently two types of aerial manipulator: Those that use an aircraft to transport one or more robotic arms, and those that incorporate several interconnected aircraft or floating elements used simultaneously for transport and handling (also called snake aerial manipulators). There are also planar aerial manipulators (which move in the plane) and omnidirectional aerial manipulators (capable of achieving independently almost any position and any spatial orientation). They also exist in rigid and flexible configurations. They can be autonomous and assisted (as a flying exoskeleton). And finally, there exists a classification of those operated by an aircraft team, by a propeller team, or by the action of valves and pumps.

Concepts linked to their use are: 

  • Omnidirectionality - the ability to provide a vehicle with independent and relatively total mobility in its translations and rotations
  • Thrust vectoring - that is the way in which the engines are distributed to achieve such omnidirectional behavior


Their main objective is to use them to manipulate objects and perform various manual tasks in the air. They are useful for the analysis and design of reconfigurable three-dimensional robots, robotics of the continuum, the basis for the understanding and the design of flying humanoids (such as the Ironman suit,) their extension to non-aerial environments (in aquatic or space applications, for example,) test benches for specialized computational or control algorithms, and test benches for novel thrust-vectoring systems.

Some global projects

Among the most outstanding projects are AIRobots, ARCAS, AEROARMS projects, the work of the Kumar-Mellinger team, the work of the Korpela-Orsag team, and the work of the Nikou-Gravidis team. More recently, we can include the DRAGON and Hydrus projects by Moju Zhao, the ODAR and LASDRA projects by D.Lee, the INROL, those of the Pucci-Fiorio team, those of the Yamaguchi-Ambe team, and of course the personal projects of Mendoza Ehecatl and Quetzalcoatl.

Current challenges

These are six of the most important challenges when dealing with aerial manipulators:

  1. Power - They require moderate to high consumption. It is necessary to provide suitable energy sources that keep the aerial manipulators flying during operating times. Examples are tethered supplies, longer-lasting batteries, solar power systems, or internal combustion systems.
  2. Vectorization & omnidirectionality - The restriction in aerial manipulators' mobility depends on their scale (dimensions and weight,) so it is necessary to develop novel methods of thrust vectoring. They need to be able to move from a manual-tool scale to an industrial one.
  3. Processing - Accomplishing all the data acquisition and implementing the necessary controls implies a need for high processing units, which must be compact, light, and low-power-consuming. These processing units are to be used onboard, as the task here is to develop better autopilots and companion computers.
  4. Measurement - Locating a flying object's position and orientation in an external environment (often with high presence of noise) is a complicated process. It can be done by computer vision, radio frequencies, laser locating, and more.
  5. Modeling & control - Since aerial manipulators are a relatively recent creation, there is lots left to do when it comes to mathematical and physical analysis of their operations.
  6. Fault tolerance - We must develop methods against engine failure situations since these can quickly become dangerous falling objects. 

Along with the suggestions indicated above (and for a feasible implementation) it is convenient to read my book Advanced Robotic Vehicles Programming: An Ardupilot and Pixhawk Approach published with Apress.

About the Author

Julio Alberto Mendoza-Mendoza is a Mexican mechatronic engineer proudly graduated from UPIITA IPN, with a Master in Advanced Technologies granted by the same institution, also he has a PhD in Computer Sciences from CIC IPN, and currently is a Postdoctoral researcher at FI UNAM, where he is developing his own line of research in aerial manipulators and derived projects. His contact and publications of global scope are available in several and prestigious publishers through the web.

This article was contributed by Julio Mendoza, co-author of Advanced Robotic Vehicles Programming: An Ardupilot and Pixhawk Approach.