Diet Estimation Example Implementation

Author: João Pedro Carvalho de Souza

Index

• Overview
• Example Overview

1. Overview

This package is an example package on how implement a ROS Action Skill implementation orientated to modularity and configurability. This examples takes into consideration ros, actionlib, C++ inheritance and polymorphisms, therefore the following prerequisites should be take place before the package study:

• ROS
• ROS Action
• C++ Inheritance
• C++ Polymorphism

2. Example Overview

This examples relies on help nutritionist choose the best food to a patient according to a sequence of heuristics which can be defined into a pipeline structure.

Diet estimation pipeline.

2.1. Setup package

2.1. Candidates dataset

The candidate.yaml is defined by patient and describes the feasible food candidate choices. One file per patient is necessary and are provided by /patient_name_candidates.yaml. The dataset file orders the candidates by suffix index, e.g., candidate_n is the n-th candidate. An example can be verified here:

2.2. Pipeline definition

Definition of the diet estimation pipeline cascade. One file per object ordering the estimation heuristics by prefix index, e.g., n_heuristic is the n-th grasp heuristic. The prefix must be one alphanumeric digit. At least one heuristic must be defined. The example bellow set a pipeline with Protein Scorer -> Fiber Sorer workflow of costs:

pipeline:
    0_protein_scorer:
        weight: 1
        threshold: 10 #in grams
    1_fiber_scorer:
        weight: 3
        threshold: 1 #in grams

2.3. Heuristics

2.3.1. Protein scorer

Select the bast candidate by amount of protein in grams. If any value is under the threshold value, the candidate is automatically removed. Parameter list:

• threshold: threshold value in grams to eliminate candidate.
• weight: a scalar value of heuristic importance. Bigger value, more this heuristic affect the overall result.

2.3.2. Fiber scorer

Select the bast candidate by amount of fiber in grams. If any value is under the threshold value, the candidate is automatically removed. Parameter list:

• threshold: threshold value in grams to eliminate candidate.
• weight: a scalar value of heuristic importance. Bigger value, more this heuristic affect the overall result.

4. How to run

In the terminal execute the server by typing:

roslaunch diet_estimation_skill_server run.launch

Call the goal to "Fulano" patient:

rostopic pub /DietEstimationSkill/goal diet_estimation_skill_msgs/DietEstimationSkillActionGoal "header:
  seq: 0
  stamp:
    secs: 0
    nsecs: 0
  frame_id: ''
goal_id:
  stamp:
    secs: 0
    nsecs: 0
  id: ''
goal:
  operation_mode: 0
  name: 'fulano'"

Check the result by:

rostopic echo /DietEstimationSkill/result