In the responses to the question “What would you do differently?”, we identify two clusters of responses: those who would like to complement their approach with the algorithms and sensors they did not use (14), and those who would like to simplify their approach (6). One team would not make any changes and four teams did not respond to the question. Among those who want to make improvements by increasing functionality, using more 3D perception and object geometry (5), using color and histogram information (2), and exploiting texture/features (2), were the common themes.