ESL UNIVERSITEIT STELLENBOSCH Electronic Systems Laboratory - - PowerPoint PPT Presentation

UNIVERSITEIT STELLENBOSCH UNIVERSITY Electronic Systems Laboratory

ESL

Neural Network Disturbance Rejection for a Quadcopter

Henry Kotz´ e Supervisors: Dr. HW Jordaan, Dr. H Kamper

Electronic System Laboratory (ESL) Department of Electrical & Electronic Engineering Stellenbosch University, South Africa

4 September 2020

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 1/13

Background

ESL

Quadcopters are being introduced in various sectors for proximity inspection, delivery and surveillance purposes. These requirements introduces phenomena not encounter by the hobbyist. A phenomenon during take-off and landing of a quadcopter is ground effects.

Inspecting walls Transporting a suspended package

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 2/13

Background

ESL

Ground effects are seen as disturbance from a control system perspective. It is omitted during the mathematical modelling and thus the control system is unaware of these dynamics. Obtaining a mathematical model is very difficult. Neural networks can be of use in this regards.

Ground Effects

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 3/13

Our Approach

ESL

The neural network assist the classical controller and does not replace it. The neural network should provide an estimate of this disturbance using the sensor measurements. This estimated disturbance can be introduced in the controller architecture to improve the disturbance rejection.

Controller

Σ

r

Σ

Plant Disturbance Observer u

Σ

Sensor Noise Disturbances Plant Output

Active disturbance rejection

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 4/13

Sim2Real

ESL

Data generation does not scale well in robotics. Training data is generated in a simulation environment. Gazebo is a physic engine to simulated rigid body dynamics. Includes realistic sensor models.

Quadcopter in Gazebo.

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 5/13

Simulated Flight

ESL

The setpoints which the quadcopter flies contains step, ramp and exponential functions. Disturbances are random pulse trains in each body direction.

Setpoint containing expected properties.

200 400 600 800 1000 Time step - [ms] −4 −2 2 4 Force or Torque - [N] or [N/m]

Random pulse train

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 6/13

Doman Randomisation

ESL

Environmental and model variables are randomised. Helps with Sim2Real tranfer. Model must learn multiple environments. Parameter Scaling factor range Additive term range mass uniform([0.95,1.05])

• principles of inertia

uniform([0.95,1.05])

• products of inertia
• uniform([0,0.0005])

gravity vector(x,y,z)

• N(0, 0.2)

Ranges of physics parameter randomisations

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 7/13

Learning to Identify Disturbances

ESL

The neural network must first learn how a quadcopter behaves without disturbances. Any deviation from the expected behaviour is the manifestation of a disturbance. By viewing the correct signals, the neural network aught to identify when a disturbance is occurring and estimating its force.

1000 2000 3000 4000 Timestep -[ms] −40 −20 20 40 Pitch angle reference -[degrees] Pitch reference Disturbance Pitch −2 −1 1 2 Force disturbance in body x-direction -[N] Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 8/13

Training Data

ESL

Input to neural network is a time window of about 0.3s Time window contains 24 vectors corresponding to the quadcopters position, velocity, acceleration and more.

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 9/13

Disturbance Neural Network

ESL

Based on OpenAI’s neural network architecture Dense ReLU layer with LSTM Optimiser: Adam

Normalised Noisy Observation

[50, 27]

Fully Connected ReLU

[1024]

LSTM

[512]

Estimated Disturbance

[3, 1]

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 10/13

Disturbance Rejection

ESL

Feeding the estimated disturbance in the correct feedback loop of the control system will counteract the disturbance much faster than originally.

Controller

Σ

r

Σ

Plant Disturbance Observer u

Σ

Sensor Noise Disturbances Plant Output

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 11/13

Responses from Disturbance Observers

ESL

500 1000 1500 2000 2500 3000 3500 4000

• 0.03
• 0.02
• 0.01

0.01 0.02 0.03 0.04 0.05 0.06 0.07

Response of quadcopter using standard control laws and using disturbance rejection with neural network.

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 12/13

Thank you

ESL

Questions? Any suggestions is appreciated

Maties Machine Learning Sept 4nd 2020 Neural Disturbance Rejection 13/13