Dual Degree Project


Autonomous Soaring of UAVs using Thermals

Abstract
A thermal is a column of warm rising air caused by differential heating on the ground. Many land birds like vultures and certain hawks use these thermals for long flights without flapping even once. Recent studies have tried using this strategy of thermal soaring for UAVs to improve their range and endurance. Many researchers have successfully flown UAVs autonomously with thermal soaring method. Autonomous thermal soaring includes many challenges like thermal detection, localization of its center, extraction of energy by circling inside the thermal, path planning to encounter maximum number of thermals, etc. The present work explores different strategies available in literature and verify them through detailed simulations. Different methods for modelling thermals are also studied to make the simulations more accurate.

Content
The above figure shows a 6 DOF simulation of an MAV entering, circling and rising inside a thermal

Objectives accomplished till now -
  • Extensive literature survey related to thermal soaring for UAVs
  • 6 DOF simulation of a glider
  • Simulation of netto variometer to detect and measure strength of a thermal
  • Implemented algorithms available in literature to estimate and track the center of the thermal
  • Implemented algorithms to circle inside a thermal

Future objectives -
  • To develop optimal control strategies so that multiple UAVs flying in a field of thermals can cooperatively exploit them to stay aloft for as long as possible
The figure below shows performance of the thermal center tracking algorithm.
The blue data points represent the x-y coordinates of the aircraft as it circles in a thermal
The red data points represent the boundary of the cylindrical thermal
The green data points represent the x-y coordinates of the predicted thermal center


For more details, you can find the copy of my DDP Stage 1 report here.


My undergraduate thesis

Multidisciplinary Design Optimization of Micro Aerial Vehicles


The above photo shows my first MAV. I spent a year modelling its subsystems, integrating them and optimizing the design before manufacturing it. Although this one was not the optimal configuration predicted by the model, it was close to it. The model predicted an endurance of 24 minutes. It flew for 19 minutes yesterday evening on a manual flight. I was elighted, primarily because 19 minutes on a 300mm fixed wing was an achievement since even the MAVs manufactured by National Aerospace Laboratories were said to clock only 25 minutes. But I later realized that I had missed the objective of the study.

MAVs, although small, are complex multidisciplinary systems and to develop optimal designs, it is necessary to model their behavior accurately. The objective of my project was not to manufacture an MAV with the highest endurance in the country but to develop a model that would accurately predict the endurance of an MAV.

Its modified version with APM 2 for autopilot


Anyways, most of the subsystems of an MAV had already been modelled. A quick glance through literature would show academic papers which provide insights into aerodynamics of various planforms shapes and sizes, interaction of propeller wash with planform aerodynamics, blade element theory for propellers, BLDC motor models, etc. All I had to do was integrate them. Two major areas relatively untouched were - 
  • Accurate prediction of battery discharge time without including complex battery chemistry 
  • Quantifying and relating quality of surveillance video with motion of MAV
These were the two fields where I contributed the most.. The entire study was condensed into a journal paper to be published in "Advances of Micro Air Vehicles in India". The paper can be viewed here.

The report submitted to the review panel of my thesis can be found here.

The study in its initial stages was also presented at the International Micro aerial Vehicle conference 2012. The conference paper along with the presentation are given here.



Chaturpaksha - Quadrotor by MAV LAB IIT Bombay


The quadrotor is named 'Chaturpaksha' which means four winged in Hindi. To know more about the timeline and budget of this project, click here.

Chaturpaksha features -

  • Size less than 300 mm
  • Total weight of ~ 600 gms
  • Payload capacity ~500 gms
  • Endurance (without payload) of 22 minutes
  • Full Carbon fiber 3 piece structure precisely cut using CNC machine
  • High quality carbon fiber plates manufactured inhouse
  • 5030 slow fly propellers
  • AXI 2203-52 race motors
  • Zippy compact 4000 mAh battery
  • 4 medium range IR sensors for collision avoidance
  • 1 long range IR sensor for altitude hold
  • 500mW 5.8 GHz video transmitter with line of sight transmission range of ~ 1 km
  • 72 MHz RC
  • 2.4 GHz Xbee for telemetry
  • ublox GPS
  • APM 2.5
  • Fat Shark FPV system
This video shows the collision avoidance capabilities for indoor navigation. This video shows some footage of the earliest model. The net was held underneath so that the battery could be fully discharged to calculate the maximum endurance

The version of Chaturpaksha shown below features a magnetic retractable arm (half retracted in the picture) for picking up and dropping light weight balls. The arm is made using a servo connected to an umbrella rod!


MAV LAB team wins third place at IMAV 2013


Cheer to MAV LAB team!. The dedication and hard work demonstrated by our team for the past few months has really paid off. Although we believe that we could have performed better, the experience we carry with ourselves is our biggest trophy.

'Chaturpaksha' 1, 2 and 3 ready to fly before the competition - 



Click here for detailed description of our unique quadrotor 'Chaturpaksha'

Below is a video clip from the event

Single propeller VTOL


This was the first idea I came up with when I decided working on 300mm, high endurance VTOLs.
It features -

  • Total weigh of ~ 500 gms
  • Payload capacity of ~300 gms
  • Endurance of 19 minutes
  • Single 1060 slowfly APC propeller
  • Structure made of chloroplast
  • Turnigy nano tech 4500 mAh battery
  • Rimfire 0.1 motor
  • MultiWii or a KK multicopter board
  • 4 control surfaces powered by turnigy 9 gram servos
The model was highly efficient due to a single motor-propeller combination and a larger diameter propeller. The main disadvantage of this configuration, the chief reason why I stopped working on it, was because it was impossible to fly it even in moderate winds due to its large surface area.

The larger version for controller testing -