If your project involves doing physical experiments, you will need to develop a plan, assemble the resources and equipment needed, recruit assistants, possibly recruit human subjects, and execute your experiment. It is of utmost importance to plan your experiments in meticulous detail and practice it before executing. Doing experiments have a higher resource cost (money, people's time, loaned equipment, etc.) than theoretical and simulation studies and you want to minimize the chances that those resources are wasted in any way. You absolutely do not want to have to redo a human subjects experiment. A well designed and written plan will make it easy for others to help you succeed.

This is general process we recommend following to ensure a successful experiment:

Develop the theoretical foundation

This often occurs during a literature study phase and kickoff phase of your project. You should develop a research question and/or hypotheses that can be answered via, in some part, experimentation.

Develop measurement specifications

Based on the mathematical theory you've developed, you will need to measure different physical quantities. It is important to identify which quantities can be measured directly (e.g. temperature sensor for temperature) and those that need to be estimated (e.g. altitude from barometric pressure). You should create minimum specifications for your measurement needs that include accuracy, precision, range, sampling rate, streaming data, time synchronization, etc.

Look for measurement solutions

Once you know precisely your minimum specifications for the desired measurements, you have to find sensors and tools to collect these measurements. You should source these from the bike lab's existing equipment, the 3mE MeetShop, borrow from other labs at the University, build the sensors yourself, or buy new sensors (in that order). To understand the possibilities, chat with existing lab members and refer to the many textbooks on engineering and scientific measurements (e.g. [Morris2001]. Developing your measurement specifications and looking for measurement solutions is an iterative process. You will need to adjust your specifications once you learn what resources are realistically available.

Design your experiments

You now need to work out what your experiment will be and how you will do it. Here you answer questions like: how many subjects do I need? how many repetitions are necessary? what variables are changed and which are fixed? where will the experiments occur? who will perform them? There is a scientific field called Design of Experiments dedicated to designing experiments that can be evaluated in a statistically sound way and provides methods on deciding variables, number of trials, repetitions, etc. Books like [Anderson2018] give an introduction to the ideas. Many of the other things to figure out at this stage are specific to the nature of the experiment and involve attending to various logistics.

Write a experimental plan

Once you know what you want to do (in a precise scientific and engineering sense), you should draft an experimental plan document. The purpose of this document is two-fold: 1) to organize your ideas and plans for yourself and 2) to communicate your plans to others. The document should include an objective for the experiments and introduction, description of the experimental methods, resources you will use and/or need, a budget, time planning, location information, safety concerns and how you can address them, and a protocol for the experiment (step-by-step recipe of the experiment). See this document for ideas about what can be in a protocol. Share this document with your advisors to get feedback and to initiate their support in obtaining resources for your experiment.

Minimal viable measurement setup

Once you have a plan you need to try out elements of your methods to make sure they actually work as you think they will. This usually involves a series of mini-tests and mini-experiments to get all of your measurements working. It may also involve creating new sensors and measurement techniques. Try to make use of existing equipment to validate you ability to measure what you want before purchasing or creating new equipment.

Pilot experiment

Once you have your measurement system working you will need to pilot your protocol. If your experiment involves human subjects this typically involves testing the experiment on yourself. You should execute your protocol completely as if it were the real experiment and collect a pilot set of data to analyze. Make sure to review all safety aspects of your protocol with your advisor(s) before performing the experiment. Take copious notes on what does and doesn't work in your protocol, so that you adjust for the real experiments.

Analyze pilot data

It is critical that you analyze your pilot data to ensure your measurements provide sufficient information to guaranteed that your study's overall data analysis can succeed. Much of the time this means ensuring that you can calculate all summary statistics reliably and accurately. You don't want to find out after doing the final experiments that you have missed a critical measurement or that a measurement is too noisy to be useful. Once again, the experiment piloting process is iterative. You have to iterative until you know your experiment functions and provides adequate data.

Finalize the protocol

Once you know your experiment is sound, write up the finalized protocol. Make sure it has all details necessary for another person to execute the experiment.

Human subjects ethics

If your study involves human subjects you need to have your experiment reviewed by the TU Delft Human Research Ethics Committee. Follow the procedures outlined by the committee. We have past examples in the lab Google Drive. You may be able to submit this before the protocol is finalized, as long as nothing would be different with respect to ethical concerns.

Prepare for the real experiments

Finalize all of your preparations, including recruiting human subjects, recruiting assistants, planning with partners, and organize all the final logistics.

Execute your experiments

If all of the above is in order, then the real experiments should go like clockwork.