Ph.D. (Engg): “Design and Development of Novel Quadcopters for Reliable Operations in Cluttered Environments”

The quadcopters are increasingly used in cluttered environments as rapid advancements are made in the development of lightweight sensors and payloads. Intelligence Surveillance Reconnaissance (ISR) missions,  crack detection on the interior surface of a pipe/tunnel, and close inspection in tropical forest environments are a handful of examples where quadcopters are being deployed. Safe operation in these cluttered environments is challenging mainly due to the proximity of obstacles to the spinning propellers. The complete loss of the propeller makes it impossible to have full-attitude stability on traditional quadcopters. After the propeller failure, the existing literature relies upon reduced-attitude control, where the control authority about yaw is sacrificed. To maintain reduced attitude control, the quadcopter must continuously spin rapidly about the yaw axis. Such a maneuver is risky, and the quadcopter may not continue the mission after the actuator fails completely.

For reliable operation in a cluttered environment, the quadcopter should also be able to reduce its span mid-flight to minimize the risk of the propeller collision with the obstacles. The quadcopter should remain fully controllable for all spans to enhance usability and applicability. The degree of span reduction should be controllable between the nominal and extreme states. Ideally, the quadcopter should also be tolerant to the complete failure of the additional “span-reducing” actuator (not to be confused with primary rotor-based actuators). For the broader range of applications, the concept or the mechanism of span-reducing should be weight-scalable.  The effective execution of an indoor cluttered environment mission may also require a mid-flight flipping quadcopter for gaining the perception of the environment along both nadir and zenith directions with respect to the payload.  Traditional quadcopters cannot sustain the inverted flight and thus lack the maneuverability and reliability to operate safely and effectively in a cluttered environment. Enhancements to the fundamental principles governing quadcopter dynamics are required to facilitate challenging operations in cluttered environments.

The first half of the presentation consists of the design and development of a morphing quadcopter called Scissorbot. Scissorbot is a novel mid-flight reconfigurable geometry quadcopter that reduces its lateral span using a single servo-motor coupled with a compact bevel differential gearbox. Scissorbot possesses unique practical features, including weight-scalability, geometrical symmetricity, and fault tolerance to the servo-motor. Scissorbot achieves significant lateral-span reduction without the risk of propeller tip collision by positioning adjacent propellers in different planes. The maximum lateral-span reduction is 88% of its nominal value (highest reported in the literature). This work derives a detailed attitude dynamics model and analyzes the gearbox theoretically. Attitude control is accomplished by implementing a Sliding Mode Controller (SMC) that exhibits robustness to parametric uncertainties such as the moment of inertia and aerodynamic disturbances due to the overlapping of the propellers. The control allocation loop is parametrized with respect to the morphing angle to adapt to the reconfiguration process.  The performance of the Scissorbot is validated using simulations, test-benches as well as real-world free-flight experiments.

The other half presents the design and development of a novel Variable-Pitch-Propeller (VPP) quadcopter called Heliquad. The cambered airfoil propeller-equipped Heliquad generates significantly more torque than its symmetrical airfoil counterpart, ensuring full-attitude hover equilibrium on only three of its working actuators. VPPs can generate reverse thrust, enabling mid-flight flip and sustained inverted flight on Heliquad. A unified control architecture ensures the tractability of the Heliquad. Furthermore, a Neural-Network (NN) based control allocation method is proposed to address the non-linearities in the actuator dynamics. The control allocation is reconfigurable based on the index of the faulty actuator. For the experimental validation, a prototype of  Heliquad is built. The design and analysis of the VPP mechanism installed on the Heliquad prototype are also presented. The performance of Heliquad is validated using simulations, test-benches, and real-world free-flight experiments. The safe recovery of a quadcopter architecture (Heliquad) with full attitude control after the complete failure of an actuator is demonstrated for the first time in the literature.

Speaker:  Kulkarni Eeshan Prashant

Research Supervisor: Prof Suresh Sundaram