Business+Impact has introduced a series on U-M alumni who have created social enterprises and continue the work of entrepreneurship after graduation. For this edition, we are shining the light on  Sea Spider, founded by Joe Huang.

Joe grew up on the ocean’s edge and developed an early awareness of ocean pollution. He initially started investigating trash cleanup in waterways and quickly discovered the larger issue of discarded fishing gear. Soon Joe has assembled a team with passionate designers and diverse skill sets: Adam Zhang, Flora Luo, Joseph Vogelpohl, and Rim Bououdina. The team started by looking to create a robotic spider that crawled on the ocean floor looking for nets — inspiring the team name.  But after several stakeholder interviews, the team eventually settled on building additional software to retrofit autonomous, underwater drones already in use with artificial intelligence that can predict when and where nets can be found. 

What inspired you to create Sea Spider?

My inspiration for founding Sea Spider is deeply rooted in my lifelong connection to the ocean. Having spent a considerable part of my childhood living near coastal regions, my perspective was profoundly influenced, particularly during my time in Guam for a year and a half. It was there that I had the privilege of exploring the Pacific Ocean through snorkeling adventures, marveling at the extraordinary marine life thriving beneath the waves. The sheer beauty and diversity of these underwater ecosystems left an indelible mark on me. In the summer of 2022, Adam and I flew to Hawaii to work with three organizations that we interviewed in the past to understand their challenge and experience that in person.

Describe what Sea Spider does.

Sea Spider specializes in enhancing the identification and removal of abandoned fishing nets in delicate coral reef ecosystems. Leveraging advanced Computer Vision technology and GPS tags, our goal is to streamline the process for conservation organizations. Our approach reduces the time spent searching for these nets, which are detrimental to marine wildlife and coral reefs. Our mission is to facilitate more effective and efficient conservation efforts in protecting these vital ecosystems.

What is your biggest recent discovery about founding your business? 

I started by trying to solve the issue of plastics around the beach. But then I realized that the ghost nets are much larger than they expected, thousands of pounds each, and require large boats and labor to remove successfully. Marine litter poses more significant threats to marine ecosystems than I imagined, and ghost nets are especially deadly to plant and animal life alike:

• 500,000 to 1 million tons of ghost nets are lost every year (Ghost Gear Report, n.d).
• 650,000 marine mammals are killed by ghost nets every year (Smith et al., 2014)
• Millions of other fish, crustaceans, and high-value species are killed yearly as a result.
• Important marine ecosystems, including coral reefs, seagrass beds, and mangrove forests are damaged as the nets travel and drag across the ocean floor.

What business school courses or UM entrepreneurship programs have helped you the most in building out the business?

I credit the University of Michigan’s Innovation in Action Competition with providing the initial foundation for this idea. It was in this program that I gained valuable insights about creating a user-centered solution. Additionally, the Michigan Business Challenge-Seigle Impact Track played a pivotal role in honing my understanding of the business aspects of this concept. Through both programs, I was challenged to assess the viability of my business and refine my vision, pushing me to consider the business’s self-sustainability. These experiences have been instrumental in shaping the trajectory of our business.

How much data on ghost nets have you gathered, and how is it stored for analysis?

Gathering a substantial dataset specifically tailored to ghost fishing nets has been a significant challenge. We initially utilized the “TrashCan 1.0” dataset, which included underwater images with various types of debris. Processing this dataset for our Box Detection and Convolutional Network approaches was demanding. We invested substantial time and effort in fine-tuning and preparing the dataset, as well as partitioning the data for Deep Learning purposes.

However, the dataset lacked available validation images, necessitating the creation of our own, using provided JSON files with image labels. This process involved extracting relevant information and generating accurate bounding boxes and binary semantic segmentation masks to train and assess our neural network.

In our quest to amass data on ghost nets, we encountered limitations. To supplement our dataset, we turned to other underwater databases and manually selected images and videos containing abandoned fishing nets. Additionally, we reached out to volunteer organizations in Hawaii to obtain further data. Our approach has encompassed Net Detection, Object Detection, Box Detection, and Deep Learning Semantic Segmentation methods, all contributing to our ongoing efforts in addressing this challenging issue.

What new ideas are you looking to pursue next, or what connections are you looking to make in the near future?

My next venture differs significantly from Sea Spider, focusing on land safety rather than ocean conservation. It addresses the pressing issue of drowsy driving, a major concern for truck drivers. Trucking accidents involving drowsy driving tend to result in more severe injuries compared to typical accidents, especially given the size of the vehicles involved. In 2021, over 100,000 drowsy driving accidents were reported in the US, with drowsy driving contributing to 40% of semi-truck crashes.

This issue arises from the demanding schedules and long working hours that truck drivers endure. They are often required to drive up to 11 hours a day, with shifts lasting 14 hours. Finding suitable overnight parking is a challenge, resulting in an average of only 5 hours of sleep per night for drivers. Due to these challenging arrangements, 64% of drivers report feeling drowsy while driving, and 18% admit to having fallen asleep at the wheel.

Addressing this problem will be a priority for my next endeavor, as it aims to enhance safety for truck drivers and reduce the risk of drowsy driving accidents.